Fossil

a {
  background-color: transparent;
  color: #ff8000;
  text-decoration: unset
}
a:active,
a:hover,




abbr[title] {
  border-bottom: 1px dotted
}
b,
optgroup,
strong,
td.usetupEditLabel {

.tl-line.warp {
  background: #600000
}
table.login_out .login_out_label {
  font-weight: 700;
  text-align: right
}

table.diff {
  width: 100%;
  overflow: auto;
  padding: 0 5px;
  font-size: 1rem;
  background: #000;
  border-radius: 5px;
}

table.diff pre {

  font-size: 1.15rem;
  scrollbar-color: black #999;
}
table.udiff pre {
  padding: 10px 0
}
td.difftxt {
  width: 52rem;

}
td.diffln ins {
  background-color: #559855;
  color: #000;
  text-decoration: none;
}
td.diffln del {
  background-color: #c55;
  color: #000;
  text-decoration: none;
}
td.difftxt del {
  background-color: inherit;
  text-decoration: none;
}
td.difftxt del > del {
  background-color: #c55;
  color: #000;
  text-decoration: none;
}
td.difftxt ins {

  background-color: inherit;
  text-decoration: none;
}
td.difftxt ins > ins {

  background-color: #559855;
  color: #000;
  text-decoration: none;
}



table.report {
  width: 100%;
  cursor: auto;
  margin: 0 0 1em;
  color: #000
}
table.report thead {

table.login_out td {
  border: 0;
}


/* Diff displays
––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– */
table.diff {
  width: 100%;
  overflow: auto;
  border: 1px solid #ccc;
  padding: 5px;
  font-size: 1rem;
}

table.diff:focus {
  outline: none;
}


/* Ticket Reports
––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– */
table.report {

  border-left: 1px dashed #bbb;
  background: rgba(255,255,255,0);
}

/************************************************************************
diffs...
************************************************************************/
td.diffln ins {
  background-color: #559855;
  color: #000;
  text-decoration: none;
}
td.diffln del {
  background-color: #c55;
  color: #000;
  text-decoration: none;
}
td.difftxt del {
  background-color: inherit;
  text-decoration: none;
}
td.difftxt del > del {
  background-color: #c55;
  color: #000;
  text-decoration: none;
}
td.difftxt ins {
  background-color: inherit;
  text-decoration: none;
}
td.difftxt ins > ins {
  background-color: #559855;
  color: #000;
  text-decoration: none;
}




/************************************************************************
************************************************************************/
body.wikiedit #fossil-status-bar,
body.fileedit #fossil-status-bar{
  border-radius: 0.25em 0.25em 0 0;
}

}

body.forum .forumPostBody > div blockquote {
    border: 1px inset;
    padding: 0 0.5em;
}





body.report table.report tr td { color: black }
body.report table.report a { color: blue }
body.tkt td.tktDspValue { color: black }
body.tkt td.tktDspValue a { color: blue }

body.branch .brlist > table > tbody > tr:hover:not(.selected),
body.branch .brlist > table > tbody > tr.selected {
  background-color: #442800;
}

.footer {
  border-top: 1px solid #ccc;
  padding: 10px;
  font-size: 0.7em;
  margin-top: 10px;
  color: #ccc;
}












/* Forum */

.forum a:visited {
  color: #6A7F94;
}

  border-top: 0px dashed #fff;
  border-left: 1px dashed #fff;
  background: rgba(255,255,255,0);
}


/* Side-by-side diff */
table.splitdiff {
  background-color: #485D7B;
  font-family: fixed, Dejavu Sans Mono, Monaco, Lucida Console, monospace;
  font-size: 8pt;
  border-collapse:collapse;
  white-space: pre;
  width: 98%;
  border: 1px #000 dashed;

/* List of files in a timeline */
ul.filelist {
  margin-top: 3px;
  line-height: 100%;
}

/* side-by-side diff display */
div.splitdiff {
  font-family: monospace;
  font-size: smaller;
  white-space: pre;
}

/* context diff display */
table.udiff {
  font-family: monospace;
  white-space: pre;
}

/* added code in a diff */
td.difftxt ins > ins, td.diffln ins {
  background-color: rgb(100, 200, 100);
}
td.difftxt ins {


  background-color: inherit;
}

/* deleted in a diff */
td.difftxt del > del, td.diffln del {
  background-color: rgb(230, 110, 110);
}


td.difftxt del {







  background-color: inherit;
}

.fileage tr:hover {
  background-color: #7EA2D9;
}

span.modpending {

  background-color: #7EA2D9;
}

.timelineModernCell[id], .timelineColumnarCell[id], .timelineDetailCell[id] {
  background-color: #455978;
}






.capsumOff {
  background-color: #bbbbbb;
}
.capsumRead {
  background-color: #006d00;
}
.capsumWrite {

/**************************************
 * Diffs
 */

/* Code Added */
td.diffln ins,
td.difftxt ins > ins {
  background-color: #7f7;
  color: #000;
}
td.difftxt ins {


  background-color: inherit;

}

/* Code Deleted */
td.diffln del,
td.difftxt del > del {
  background-color: #f77;
  color: #000;
}
td.difftxt del {
  background-color: inherit;
}


/**************************************
 * Diffs : Side-By-Side
 */

/* display (column-based) */
table.splitdiff {
  border-spacing: 0;
  font-size: 0.85rem;
  width: 90%;
}

table.splitdiff pre {
  border: 0;
  margin: 0 0.5em;
  padding: 0;
}

table.splitdiff td {
  padding: 0;
  vertical-align: top;
}

/* line number column */
td.diffln {
  color: #ee0;
  padding-right: 0.75em;
  text-align: right;
}

/* diff text column */
td.difftxt {
  background-color: #111;
  overflow-x: auto;
  width: 45em;
}








/* diff marker column */
td.diffsep {
  padding: 0 0.5em;
}


/**************************************
 * Diffs : Unified
 */

table.udiff pre {
  background-color: #111;
}








/**************************************
 * File List : Flat
 */

table.browser {
  width: 100%;

** Copy a blob
*/
void blob_copy(Blob *pTo, Blob *pFrom){
  blob_is_init(pFrom);
  blob_zero(pTo);
  blob_append(pTo, blob_buffer(pFrom), blob_size(pFrom));
}

/*
** Append the second blob onto the end of the first blob and reset the
** second blob.
*/
void blob_append_xfer(Blob *pTo, Blob *pFrom){
  blob_append(pTo, blob_buffer(pFrom), blob_size(pFrom));
  blob_reset(pFrom);
}

/*
** Write into pOut, a string literal representation for the first n bytes
** of z[].  The string literal representation is compatible with C, TCL,
** and JSON.  Double-quotes are added to both ends.  Double-quote and
** backslash characters are escaped.
*/
void blob_append_tcl_literal(Blob *pOut, const char *z, int n){
  int i;
  blob_append_char(pOut, '"');
  for(i=0; i<n; i++){
    char c = z[i];
    switch( c ){
      case '\r':  c = 'r';
      case '[':
      case ']':
      case '$':
      case '"':
      case '\\':
        blob_append_char(pOut, '\\');
      default:
        blob_append_char(pOut, c);
    }
  }
  blob_append_char(pOut, '"');
}
void blob_append_json_literal(Blob *pOut, const char *z, int n){
  int i;
  blob_append_char(pOut, '"');
  for(i=0; i<n; i++){
    char c = z[i];
    switch( c ){
      case '"':
      case '\\':
        blob_append_char(pOut, '\\');
      default:
        blob_append_char(pOut, c);
    }
  }
  blob_append_char(pOut, '"');
}


/*
** Return a pointer to a null-terminated string for a blob.
*/
char *blob_str(Blob *p){
  blob_is_init(p);
  if( p->nUsed==0 ){

ul.filelist {
  margin-top: 3px;
  line-height: 100%;
}
ul.filelist li {
  padding-top: 1px;
}

/* Rules governing diff layout and colors */
table.diff {
  width: 98%;
  border-spacing: 0;

  border: 1px solid black;
}
table.diff td {

  vertical-align: top;
}
table.diff pre {
  margin: 0 0 0 0;





}
td.diffln {
  width: 1px;
  text-align: right;
  padding: 0 1em 0 0;
}
td.difflne {
  padding-bottom: 0.4em;
}
td.diffsep {
  width: 1px;
  padding: 0 0.3em 0 1em;
}
td.difftxt pre {
  overflow-x: auto;
}
td.diffln ins {
  background-color: #a0e4b2;
  text-decoration: none;
}
td.diffln del {
  background-color: #ffc0c0;
  text-decoration: none;
}
td.difftxt del {
  background-color: #ffe8e8;
  text-decoration: none;
}
td.difftxt del > del {
  background-color: #ffc0c0;
  text-decoration: none;
  font-weight: bold;
}
td.difftxt del > del.edit {
  background-color: #c0c0ff;
  text-decoration: none;
  font-weight: bold;
}
td.difftxt ins {
  background-color: #dafbe1;
  text-decoration: none;
}
td.difftxt ins > ins {
  background-color: #a0e4b2;
  text-decoration: none;
  font-weight: bold;
}
td.difftxt ins > ins.edit {
  background-color: #c0c0ff;
  text-decoration: none;
  font-weight: bold;
}


span.modpending {
  color: #b03800;
  font-style: italic;
}
pre.th1result {
  white-space: pre-wrap;
  word-wrap: break-word;

#define DIFF_NUMSTAT      ((u64)0x80000000) /* Show line count of changes */
#define DIFF_NOOPT        (((u64)0x01)<<32) /* Suppress optimizations (debug) */
#define DIFF_INVERT       (((u64)0x02)<<32) /* Invert the diff (debug) */
#define DIFF_CONTEXT_EX   (((u64)0x04)<<32) /* Use context even if zero */
#define DIFF_NOTTOOBIG    (((u64)0x08)<<32) /* Only display if not too big */
#define DIFF_STRIP_EOLCR  (((u64)0x10)<<32) /* Strip trailing CR */
#define DIFF_SLOW_SBS     (((u64)0x20)<<32) /* Better but slower side-by-side */
#define DIFF_WEBPAGE   (((u64)0x00040)<<32) /* Complete webpage */
#define DIFF_BROWSER   (((u64)0x00080)<<32) /* The --browser option */
#define DIFF_JSON      (((u64)0x00100)<<32) /* JSON output */
#define DIFF_DEBUG     (((u64)0x00200)<<32) /* Debugging diff output */
#define DIFF_RAW       (((u64)0x00400)<<32) /* Raw triples - for debugging */
#define DIFF_TCL       (((u64)0x00800)<<32) /* For the --tk option */

/*
** These error messages are shared in multiple locations.  They are defined
** here for consistency.
*/
#define DIFF_CANNOT_COMPUTE_BINARY \
    "cannot compute difference between binary files\n"

};

/*
** Length of a dline
*/
#define LENGTH(X)   ((X)->n)

/*
** Number of diff chunks generated
*/
static int nChunk = 0;

/*
** A context for running a raw diff.
**
** The aEdit[] array describes the raw diff.  Each triple of integers in
** aEdit[] means:
**
**   (1) COPY:   Number of lines aFrom and aTo have in common

}

/*
** Return true if the regular expression *pRe matches any of the
** N dlines
*/
static int re_dline_match(
  ReCompiled *pRe,      /* The regular expression to be matched */
  const DLine *aDLine,  /* First of N DLines to compare against */
  int N                 /* Number of DLines to check */
){
  while( N-- ){
    if( re_match(pRe, (const unsigned char *)aDLine->z, LENGTH(aDLine)) ){
      return 1;
    }
    aDLine++;
  }
  return 0;
}

/*
** Append a single line of context-diff output to pOut.
*/
static void appendDiffLine(
  Blob *pOut,         /* Where to write the line of output */
  char cPrefix,       /* One of " ", "+",  or "-" */
  DLine *pLine        /* The line to be output */


){
  blob_append_char(pOut, cPrefix);













  blob_append(pOut, pLine->z, pLine->n);

  blob_append_char(pOut, '\n');
}

/*
** Add two line numbers to the beginning of an output line for a context
** diff.  One or the other of the two numbers might be zero, which means
** to leave that number field blank.

*/
static void appendDiffLineno(Blob *pOut, int lnA, int lnB){

  if( lnA>0 ){
    blob_appendf(pOut, "%6d ", lnA);
  }else{
    blob_append(pOut, "       ", 7);
  }
  if( lnB>0 ){
    blob_appendf(pOut, "%6d  ", lnB);
  }else{
    blob_append(pOut, "        ", 8);
  }

}

/*

** Output a patch-style text diff.
*/
static void contextDiff(
  DContext *p,      /* The difference */
  Blob *pOut,       /* Output a context diff to here */

  u64 diffFlags     /* Flags controlling the diff format */
){
  DLine *A;     /* Left side of the diff */
  DLine *B;     /* Right side of the diff */
  int a = 0;    /* Index of next line in A[] */
  int b = 0;    /* Index of next line in B[] */
  int *R;       /* Array of COPY/DELETE/INSERT triples */
  int r;        /* Index into R[] */
  int nr;       /* Number of COPY/DELETE/INSERT triples to process */
  int mxr;      /* Maximum value for r */
  int na, nb;   /* Number of lines shown from A and B */
  int i, j;     /* Loop counters */
  int m;        /* Number of lines to output */
  int skip;     /* Number of lines to skip */
  static int nChunk = 0;  /* Number of diff chunks seen so far */
  int nContext;    /* Number of lines of context */
  int showLn;      /* Show line numbers */

  int showDivider = 0;  /* True to show the divider between diff blocks */

  nContext = diff_context_lines(diffFlags);
  showLn = (diffFlags & DIFF_LINENO)!=0;


  A = p->aFrom;
  B = p->aTo;
  R = p->aEdit;
  mxr = p->nEdit;
  while( mxr>2 && R[mxr-1]==0 && R[mxr-2]==0 ){ mxr -= 3; }
  for(r=0; r<mxr; r += 3*nr){
    /* Figure out how many triples to show in a single block */
    for(nr=1; R[r+nr*3]>0 && R[r+nr*3]<nContext*2; nr++){}
    /* printf("r=%d nr=%d\n", r, nr); */


























    /* For the current block comprising nr triples, figure out
    ** how many lines of A and B are to be displayed
    */
    if( R[r]>nContext ){
      na = nb = nContext;
      skip = R[r] - nContext;
    }else{

    ** the previous block.
    */
    nChunk++;
    if( showLn ){
      if( !showDivider ){
        /* Do not show a top divider */
        showDivider = 1;


      }else{
        blob_appendf(pOut, "%.80c\n", '.');
      }

    }else{

      /*
       * If the patch changes an empty file or results in an empty file,
       * the block header must use 0,0 as position indicator and not 1,0.
       * Otherwise, patch would be confused and may reject the diff.
       */
      blob_appendf(pOut,"@@ -%d,%d +%d,%d @@",
        na ? a+skip+1 : a+skip, na,
        nb ? b+skip+1 : b+skip, nb);

      blob_append(pOut, "\n", 1);
    }

    /* Show the initial common area */
    a += skip;
    b += skip;
    m = R[r] - skip;
    for(j=0; j<m; j++){
      if( showLn ) appendDiffLineno(pOut, a+j+1, b+j+1);
      appendDiffLine(pOut, ' ', &A[a+j]);
    }
    a += m;
    b += m;

    /* Show the differences */
    for(i=0; i<nr; i++){
      m = R[r+i*3+1];
      for(j=0; j<m; j++){
        if( showLn ) appendDiffLineno(pOut, a+j+1, 0);
        appendDiffLine(pOut, '-', &A[a+j]);
      }
      a += m;
      m = R[r+i*3+2];
      for(j=0; j<m; j++){
        if( showLn ) appendDiffLineno(pOut, 0, b+j+1);
        appendDiffLine(pOut, '+', &B[b+j]);
      }
      b += m;
      if( i<nr-1 ){
        m = R[r+i*3+3];
        for(j=0; j<m; j++){
          if( showLn ) appendDiffLineno(pOut, a+j+1, b+j+1);
          appendDiffLine(pOut, ' ', &A[a+j]);
        }
        b += m;
        a += m;
      }
    }

    /* Show the final common area */
    assert( nr==i );
    m = R[r+nr*3];
    if( m>nContext ) m = nContext;
    for(j=0; j<m; j++){
      if( showLn ) appendDiffLineno(pOut, a+j+1, b+j+1);
      appendDiffLine(pOut, ' ', &A[a+j]);
    }
  }

}

















#define MX_CSN  8  /* Maximum number of change spans across a change region */




















































































































/*
** A description of zero or more (up to MX_CSN) areas of difference
** between two lines of text.
*/
typedef struct LineChange LineChange;
struct LineChange {
  int n;            /* Number of change spans */

  struct Span {




    int iStart1;    /* Byte offset to start of a change on the left */

    int iLen1;      /* Length of the left change in bytes */




    int iStart2;    /* Byte offset to start of a change on the right */










    int iLen2;      /* Length of the change on the right in bytes */
    int isMin;      /* True if this change is known to have no useful subdivs */
  } a[MX_CSN];     /* Array of change spans, sorted order */
};


/*
** The two text segments zLeft and zRight are known to be different on
** both ends, but they might have  a common segment in the middle.  If
** they do not have a common segment, return 0.  If they do have a large
** common segment, return 1 and before doing so set:
**

static int textLCS(
  const char *zLeft,  int nA,       /* String on the left */
  const char *zRight,  int nB,      /* String on the right */
  int *aLCS                         /* Identify bounds of LCS here */
){
  const unsigned char *zA = (const unsigned char*)zLeft;    /* left string */
  const unsigned char *zB = (const unsigned char*)zRight;   /* right string */
  int i, j, k;               /* Loop counters */
  int lenBest = 0;           /* Match length to beat */

  for(i=0; i<nA-lenBest; i++){
    unsigned char cA = zA[i];

    if( (cA&0xc0)==0x80 ) continue;
    for(j=0; j<nB-lenBest; j++ ){
      if( zB[j]==cA ){
        for(k=1; j+k<nB && i+k<nA && zB[j+k]==zA[i+k]; k++){}
        while( (zB[j+k]&0xc0)==0x80 ){ k--; }
        if( k>lenBest ){
          lenBest = k;
          aLCS[0] = i;
          aLCS[1] = i+k;
          aLCS[2] = j;
          aLCS[3] = j+k;
        }
      }
    }
  }
  return lenBest>0;
}

/*

** Find the smallest spans that are different between two text strings that
** are known to be different on both ends.
*/
static int textLineChanges(
  const char *zLeft,  int nA,       /* String on the left */
  const char *zRight,  int nB,      /* String on the right */
  LineChange *p                     /* Write results here */
){
  p->n = 1;

  p->a[0].iStart1 = 0;
  p->a[0].iLen1 = nA;
  p->a[0].iStart2 = 0;
  p->a[0].iLen2 = nB;

  p->a[0].isMin = 0;

  while( p->n<MX_CSN-1 ){
    int mxi = -1;

    int mxLen = -1;
    int x, i;
    int aLCS[4];
    struct Span *a, *b;
    for(i=0; i<p->n; i++){
      if( p->a[i].isMin ) continue;
      x = p->a[i].iLen1;
      if( p->a[i].iLen2<x ) x = p->a[i].iLen2;




      if( x>mxLen ){
        mxLen = x;



        mxi = i;
      }
    }
    if( mxLen<6 ) break;
    x = textLCS(zLeft + p->a[mxi].iStart1, p->a[mxi].iLen1,
                zRight + p->a[mxi].iStart2, p->a[mxi].iLen2, aLCS);
    if( x==0 ){
      p->a[mxi].isMin = 1;
      continue;
    }
    a = p->a+mxi;
    b = a+1;
    if( mxi<p->n-1 ){
      memmove(b+1, b, sizeof(*b)*(p->n-mxi-1));
    }
    p->n++;
    b->iStart1 = a->iStart1 + aLCS[1];
    b->iLen1 = a->iLen1 - aLCS[1];
    a->iLen1 = aLCS[0];
    b->iStart2 = a->iStart2 + aLCS[3];
    b->iLen2 = a->iLen2 - aLCS[3];
    a->iLen2 = aLCS[2];
    b->isMin = 0;
  }
  return p->n;
}

/*
** Return true if the string starts with n spaces


*/
static int allSpaces(const char *z, int n){
  int i;
  for(i=0; i<n && fossil_isspace(z[i]); i++){}

  return i==n;


}


/*

** Try to improve the human-readability of the LineChange p.
**
** (1)  If the first change span shows a change of indentation, try to
**      move that indentation change to the left margin.

**
** (2)  Try to shift changes so that they begin or end with a space.
*/
static void improveReadability(
  const char *zA,  /* Left line of the change */
  const char *zB,  /* Right line of the change */
  LineChange *p    /* The LineChange to be adjusted */
){
  int j, n, len;
  if( p->n<1 ) return;

  /* (1) Attempt to move indentation changes to the left margin */
  if( p->a[0].iLen1==0
   && (len = p->a[0].iLen2)>0
   && (j = p->a[0].iStart2)>0
   && zB[0]==zB[j]
   && allSpaces(zB, j)
  ){
    for(n=1; n<len && n<j && zB[j]==zB[j+n]; n++){}
    if( n<len ){
      memmove(&p->a[1], &p->a[0], sizeof(p->a[0])*p->n);
      p->n++;
      p->a[0] = p->a[1];
      p->a[1].iStart2 += n;
      p->a[1].iLen2 -= n;
      p->a[0].iLen2 = n;
    }
    p->a[0].iStart1 = 0;
    p->a[0].iStart2 = 0;
  }else
  if( p->a[0].iLen2==0
   && (len = p->a[0].iLen1)>0
   && (j = p->a[0].iStart1)>0
   && zA[0]==zA[j]
   && allSpaces(zA, j)
  ){
    for(n=1; n<len && n<j && zA[j]==zA[j+n]; n++){}
    if( n<len ){
      memmove(&p->a[1], &p->a[0], sizeof(p->a[0])*p->n);
      p->n++;
      p->a[0] = p->a[1];
      p->a[1].iStart1 += n;
      p->a[1].iLen1 -= n;
      p->a[0].iLen1 = n;
    }
    p->a[0].iStart1 = 0;
    p->a[0].iStart2 = 0;
  }



  /* (2) Try to shift changes so that they begin or end with a
  ** space.  (TBD) */
}


/*

** Given two lines of text, pFrom and pTo, compute a set of changes
** between those two lines, for enhanced display purposes.
**
** The result is written into the LineChange object given by the
** third parameter.
*/
static void oneLineChange(

  const DLine *pLeft,  /* Left line of the change */

  const DLine *pRight, /* Right line of the change */

  LineChange *p        /* OUTPUT: Write the results here */
){
  int nLeft;           /* Length of left line in bytes */
  int nRight;          /* Length of right line in bytes */
  int nShort;          /* Shortest of left and right */
  int nPrefix;         /* Length of common prefix */
  int nSuffix;         /* Length of common suffix */
  int nCommon;         /* Total byte length of suffix and prefix */
  const char *zLeft;   /* Text of the left line */
  const char *zRight;  /* Text of the right line */
  int nLeftDiff;       /* nLeft - nPrefix - nSuffix */
  int nRightDiff;      /* nRight - nPrefix - nSuffix */





  nLeft = pLeft->n;
  zLeft = pLeft->z;
  nRight = pRight->n;
  zRight = pRight->z;
  nShort = nLeft<nRight ? nLeft : nRight;

  nPrefix = 0;
  while( nPrefix<nShort && zLeft[nPrefix]==zRight[nPrefix] ){
    nPrefix++;
  }
  if( nPrefix<nShort ){
    while( nPrefix>0 && (zLeft[nPrefix]&0xc0)==0x80 ) nPrefix--;
  }
  nSuffix = 0;
  if( nPrefix<nShort ){
    while( nSuffix<nShort
           && zLeft[nLeft-nSuffix-1]==zRight[nRight-nSuffix-1] ){
      nSuffix++;
    }
    if( nSuffix<nShort ){
      while( nSuffix>0 && (zLeft[nLeft-nSuffix]&0xc0)==0x80 ) nSuffix--;
    }
    if( nSuffix==nLeft || nSuffix==nRight ) nPrefix = 0;
  }
  nCommon = nPrefix + nSuffix;

  /* If the prefix and suffix overlap, that means that we are dealing with
  ** a pure insertion or deletion of text that can have multiple alignments.
  ** Try to find an alignment to begins and ends on whitespace, or on
  ** punctuation, rather than in the middle of a name or number.
  */
  if( nCommon > nShort ){
    int iBest = -1;
    int iBestVal = -1;
    int i;
    int nLong = nLeft<nRight ? nRight : nLeft;
    int nGap = nLong - nShort;
    for(i=nShort-nSuffix; i<=nPrefix; i++){
       int iVal = 0;

       if( iVal>iBestVal ){
         iBestVal = iVal;
         iBest = i;
       }
    }
    nPrefix = iBest;
    nSuffix = nShort - nPrefix;
    nCommon = nPrefix + nSuffix;
  }

  /* A single chunk of text inserted */
  if( nCommon==nLeft ){

    p->n = 1;
    p->a[0].iStart1 = nPrefix;
    p->a[0].iLen1 = 0;






    p->a[0].iStart2 = nPrefix;
    p->a[0].iLen2 = nRight - nCommon;

    improveReadability(zLeft, zRight, p);
    return;
  }

  /* A single chunk of text deleted */
  if( nCommon==nRight ){


    p->n = 1;
    p->a[0].iStart1 = nPrefix;
    p->a[0].iLen1 = nLeft - nCommon;
    p->a[0].iStart2 = nPrefix;



    p->a[0].iLen2 = 0;
    improveReadability(zLeft, zRight, p);
    return;
  }

  /* At this point we know that there is a chunk of text that has
  ** changed between the left and the right.  Check to see if there
  ** is a large unchanged section in the middle of that changed block.
  */
  nLeftDiff = nLeft - nCommon;
  nRightDiff = nRight - nCommon;

  if( nLeftDiff >= 4
   && nRightDiff >= 4
   && textLineChanges(&zLeft[nPrefix], nLeftDiff,
                      &zRight[nPrefix], nRightDiff, p)>1
  ){

    int i;



    for(i=0; i<p->n; i++){










      p->a[i].iStart1 += nPrefix;
      p->a[i].iStart2 += nPrefix;





    }



    improveReadability(zLeft, zRight, p);

    return;
  }

  /* If all else fails, show a single big change between left and right */

  p->n = 1;
  p->a[0].iStart1 = nPrefix;
  p->a[0].iLen1 = nLeft - nCommon;
  p->a[0].iStart2 = nPrefix;
  p->a[0].iLen2 = nRight - nCommon;
  improveReadability(zLeft, zRight, p);
}

/*
** COMMAND: test-line-diff
** Usage: %fossil% test-line-diff STRING1 STRING2
**
** Show the differences between the two strings.  Used for testing
** the oneLineChange() routine in the diff logic.
*/
void test_line_diff(void){
  DLine a, b;
  LineChange chng;
  int i, j, x;
  if( g.argc!=4 ) usage("STRING1 STRING2");
  a.z = g.argv[2];

  a.n = (int)strlen(a.z);
  b.z = g.argv[3];
  b.n = (int)strlen(b.z);
  oneLineChange(&a, &b, &chng);
  fossil_print("left:  [%s]\n", a.z);
  for(i=x=0; i<chng.n; i++){
    int ofst = chng.a[i].iStart1;
    int len = chng.a[i].iLen1;
    if( len ){
      if( x==0 ){ fossil_print("%*s", 8, ""); }
      while( ofst > x ){
        if( (a.z[x]&0xc0)!=0x80 ) fossil_print(" ");
        x++;
      }
      for(j=0; j<len; j++, x++){
        if( (a.z[x]&0xc0)!=0x80 ) fossil_print("%d",i);
      }
    }
  }
  if( x ) fossil_print("\n");
  fossil_print("right: [%s]\n", b.z);
  for(i=x=0; i<chng.n; i++){
    int ofst = chng.a[i].iStart2;
    int len = chng.a[i].iLen2;
    if( len ){
      if( x==0 ){ fossil_print("%*s", 8, ""); }
      while( ofst > x ){
        if( (b.z[x]&0xc0)!=0x80 ) fossil_print(" ");
        x++;
      }
      for(j=0; j<len; j++, x++){
        if( (b.z[x]&0xc0)!=0x80 ) fossil_print("%d",i);
      }
    }
  }
  if( x ) fossil_print("\n");
}

/*
** Minimum of two values
*/
static int minInt(int a, int b){ return a<b ? a : b; }



/*
** This is an abstract superclass for an object that accepts difference
** lines and formats them for display.  Subclasses of this object format
** the diff output in different ways.
**
** To subclass, create an instance of the DiffBuilder object and fill
** in appropriate method implementations.
*/ 
typedef struct DiffBuilder DiffBuilder;
struct DiffBuilder {
  void (*xSkip)(DiffBuilder*, unsigned int, int);
  void (*xCommon)(DiffBuilder*,const DLine*);
  void (*xInsert)(DiffBuilder*,const DLine*);
  void (*xDelete)(DiffBuilder*,const DLine*);
  void (*xReplace)(DiffBuilder*,const DLine*, const DLine*);
  void (*xEdit)(DiffBuilder*,const DLine*,const DLine*);
  void (*xEnd)(DiffBuilder*);
  unsigned int lnLeft;              /* Lines seen on the left (delete) side */
  unsigned int lnRight;             /* Lines seen on the right (insert) side */
  unsigned int nPending;            /* Number of pending lines */
  int eState;                       /* State of the output */
  int width;                        /* Display width */
  Blob *pOut;                       /* Output blob */
  Blob aCol[5];                     /* Holding blobs */
};

/************************* DiffBuilderDebug ********************************/
/* This version of DiffBuilder is used for debugging the diff and diff
** diff formatter logic.  It is accessed using the (undocumented) --debug
** option to the diff command.  The output is human-readable text that
** describes the various method calls that are invoked agains the DiffBuilder
** object.
*/
static void dfdebugSkip(DiffBuilder *p, unsigned int n, int isFinal){
  blob_appendf(p->pOut, "SKIP %d (%d..%d left and %d..%d right)%s\n",
                n, p->lnLeft+1, p->lnLeft+n, p->lnRight+1, p->lnRight+n,
                isFinal ? " FINAL" : "");
  p->lnLeft += n;
  p->lnRight += n;
}
static void dfdebugCommon(DiffBuilder *p, const DLine *pLine){
  p->lnLeft++;
  p->lnRight++;
  blob_appendf(p->pOut, "COMMON  %8u %8u %.*s\n",
      p->lnLeft, p->lnRight, (int)pLine->n, pLine->z);
}
static void dfdebugInsert(DiffBuilder *p, const DLine *pLine){
  p->lnRight++;
  blob_appendf(p->pOut, "INSERT           %8d %.*s\n",
      p->lnRight, (int)pLine->n, pLine->z);
}
static void dfdebugDelete(DiffBuilder *p, const DLine *pLine){
  p->lnLeft++;
  blob_appendf(p->pOut, "DELETE   %8u          %.*s\n",
      p->lnLeft, (int)pLine->n, pLine->z);
}
static void dfdebugReplace(DiffBuilder *p, const DLine *pX, const DLine *pY){
  p->lnLeft++;
  p->lnRight++;
  blob_appendf(p->pOut, "REPLACE %8u          %.*s\n",
      p->lnLeft, (int)pX->n, pX->z);
  blob_appendf(p->pOut, "            %8u %.*s\n",
      p->lnRight, (int)pY->n, pY->z);
}
static void dfdebugEdit(DiffBuilder *p, const DLine *pX, const DLine *pY){
  int i, j;
  int x;
  LineChange chng;
  p->lnLeft++;
  p->lnRight++;
  blob_appendf(p->pOut, "EDIT     %8u          %.*s\n",
               p->lnLeft, (int)pX->n, pX->z);
  oneLineChange(pX, pY, &chng);
  for(i=x=0; i<chng.n; i++){
    int ofst = chng.a[i].iStart1;
    int len = chng.a[i].iLen1;
    if( len ){
      if( x==0 ){ blob_appendf(p->pOut, "%*s", 25, ""); }
      while( ofst > x ){
        if( (pX->z[x]&0xc0)!=0x80 ) blob_append_char(p->pOut, ' ');
        x++;
      }
      for(j=0; j<len; j++, x++){
        if( (pX->z[x]&0xc0)!=0x80 ) blob_append_char(p->pOut, '^');
      }
    }
  }
  if( x ) blob_append_char(p->pOut, '\n');
  blob_appendf(p->pOut, "                %8u %.*s\n",
               p->lnRight, (int)pY->n, pY->z);
  for(i=x=0; i<chng.n; i++){
    int ofst = chng.a[i].iStart2;
    int len = chng.a[i].iLen2;
    if( len ){
      if( x==0 ){ blob_appendf(p->pOut, "%*s", 25, ""); }
      while( ofst > x ){
        if( (pY->z[x]&0xc0)!=0x80 ) blob_append_char(p->pOut, ' ');
        x++;
      }
      for(j=0; j<len; j++, x++){
        if( (pY->z[x]&0xc0)!=0x80 ) blob_append_char(p->pOut, '^');
      }
    }
  }
  if( x ) blob_append_char(p->pOut, '\n');
}
static void dfdebugEnd(DiffBuilder *p){
  blob_appendf(p->pOut, "END with %u lines left and %u lines right\n",
     p->lnLeft, p->lnRight);
  fossil_free(p);
}
static DiffBuilder *dfdebugNew(Blob *pOut){
  DiffBuilder *p = fossil_malloc(sizeof(*p));
  p->xSkip = dfdebugSkip;
  p->xCommon = dfdebugCommon;
  p->xInsert = dfdebugInsert;
  p->xDelete = dfdebugDelete;
  p->xReplace = dfdebugReplace;
  p->xEdit = dfdebugEdit;
  p->xEnd = dfdebugEnd;
  p->lnLeft = p->lnRight = 0;
  p->pOut = pOut;
  return p;
}

/************************* DiffBuilderTcl ********************************/
/*
** This formatter outputs a description of the diff formatted as TCL, for
** use by the --tk option to "diff".   See also the "diff.tcl" file.  The
** output can be viewed directly using the --tcl option.
**
** There is one line per method call:
**
**     SKIP n                      -- Skip "n" lines of input
**     COM string                  -- "string" is an unchanged context line
**     INS string                  -- "string" is in the right file only
**     DEL string                  -- "string" is in the left file only
**     EDIT string ....            -- Complex edit between left and right
**
** The EDIT verb will be followed by 3*N or 3*N+1 strings.  The triples
** each show:
**
**     1.  Common text
**     2.  Text from the left side
**     3.  Text on the right that replaces (2) from the left
**
** For inserted text (2) will be an empty string.  For deleted text, (3)
** will be an empty string.  (1) might be empty for the first triple if
** the line begins with an edit.  After all triples, there might be one
** additional string which is a common suffix.
*/
static void dftclSkip(DiffBuilder *p, unsigned int n, int isFinal){
  blob_appendf(p->pOut, "SKIP %u\n", n);
}
static void dftclCommon(DiffBuilder *p, const DLine *pLine){
  blob_appendf(p->pOut, "COM ");
  blob_append_tcl_literal(p->pOut, pLine->z, pLine->n);
  blob_append_char(p->pOut, '\n');
}
static void dftclInsert(DiffBuilder *p, const DLine *pLine){
  blob_append(p->pOut, "INS ", -1);
  blob_append_tcl_literal(p->pOut, pLine->z, pLine->n);
  blob_append_char(p->pOut, '\n');
}
static void dftclDelete(DiffBuilder *p, const DLine *pLine){
  blob_append(p->pOut, "DEL ", -1);
  blob_append_tcl_literal(p->pOut, pLine->z, pLine->n);
  blob_append_char(p->pOut, '\n');
}
static void dftclReplace(DiffBuilder *p, const DLine *pX, const DLine *pY){
  blob_append(p->pOut, "EDIT \"\" ", -1);
  blob_append_tcl_literal(p->pOut, pX->z, pX->n);
  blob_append_char(p->pOut, ' ');
  blob_append_tcl_literal(p->pOut, pY->z, pY->n);
  blob_append_char(p->pOut, '\n');
}
static void dftclEdit(DiffBuilder *p, const DLine *pX, const DLine *pY){
  int i, x;
  LineChange chng;
  blob_append(p->pOut, "EDIT", 4);
  oneLineChange(pX, pY, &chng);
  for(i=x=0; i<chng.n; i++){
    blob_append_char(p->pOut, ' ');
    blob_append_tcl_literal(p->pOut, pX->z + x, chng.a[i].iStart1 - x);
    x = chng.a[i].iStart1;
    blob_append_char(p->pOut, ' ');
    blob_append_tcl_literal(p->pOut, pX->z + x, chng.a[i].iLen1);
    x += chng.a[i].iLen1;
    blob_append_char(p->pOut, ' ');
    blob_append_tcl_literal(p->pOut, 
                         pY->z + chng.a[i].iStart2, chng.a[i].iLen2);
  }
  if( x<pX->n ){
    blob_append_char(p->pOut, ' ');
    blob_append_tcl_literal(p->pOut, pX->z + x, pX->n - x);
  }
  blob_append_char(p->pOut, '\n');
}
static void dftclEnd(DiffBuilder *p){
  fossil_free(p);
}
static DiffBuilder *dftclNew(Blob *pOut){
  DiffBuilder *p = fossil_malloc(sizeof(*p));
  p->xSkip = dftclSkip;
  p->xCommon = dftclCommon;
  p->xInsert = dftclInsert;
  p->xDelete = dftclDelete;
  p->xReplace = dftclReplace;
  p->xEdit = dftclEdit;
  p->xEnd = dftclEnd;
  p->pOut = pOut;
  return p;
}

/************************* DiffBuilderJson ********************************/
/*
** This formatter generates a JSON array that describes the difference.
**
** The Json array consists of integer opcodes with each opcode followed
** by zero or more arguments:
**
**   Syntax        Mnemonic    Description
**   -----------   --------    --------------------------
**   0             END         This is the end of the diff
**   1  INTEGER    SKIP        Skip N lines from both files
**   2  STRING     COMMON      The line show by STRING is in both files
**   3  STRING     INSERT      The line STRING is in only the right file
**   4  STRING     DELETE      The STRING line is in only the left file
**   5  SUBARRAY   EDIT        One line is different on left and right.
**
** The SUBARRAY is an array of 3*N+1 strings with N>=0.  The triples
** represent common-text, left-text, and right-text.  The last string
** in SUBARRAY is the common-suffix.  Any string can be empty if it does
** not apply.
*/
static void dfjsonSkip(DiffBuilder *p, unsigned int n, int isFinal){
  blob_appendf(p->pOut, "1,%u,\n", n);
}
static void dfjsonCommon(DiffBuilder *p, const DLine *pLine){
  blob_append(p->pOut, "2,",2);
  blob_append_json_literal(p->pOut, pLine->z, (int)pLine->n);
  blob_append(p->pOut, ",\n",2);
}
static void dfjsonInsert(DiffBuilder *p, const DLine *pLine){
  blob_append(p->pOut, "3,",2);
  blob_append_json_literal(p->pOut, pLine->z, (int)pLine->n);
  blob_append(p->pOut, ",\n",2);
}
static void dfjsonDelete(DiffBuilder *p, const DLine *pLine){
  blob_append(p->pOut, "4,",2);
  blob_append_json_literal(p->pOut, pLine->z, (int)pLine->n);
  blob_append(p->pOut, ",\n",2);
}
static void dfjsonReplace(DiffBuilder *p, const DLine *pX, const DLine *pY){
  blob_append(p->pOut, "5,[\"\",",-1);
  blob_append_json_literal(p->pOut, pX->z, (int)pX->n);
  blob_append(p->pOut, ",",1);
  blob_append_json_literal(p->pOut, pY->z, (int)pY->n);
  blob_append(p->pOut, ",\"\"],\n",-1);
}
static void dfjsonEdit(DiffBuilder *p, const DLine *pX, const DLine *pY){
  int i, x;
  LineChange chng;
  blob_append(p->pOut, "5,[", 3);
  oneLineChange(pX, pY, &chng);
  for(i=x=0; i<chng.n; i++){
    blob_append_json_literal(p->pOut, pX->z + x, chng.a[i].iStart1 - x);
    x = chng.a[i].iStart1;
    blob_append_char(p->pOut, ',');
    blob_append_json_literal(p->pOut, pX->z + x, chng.a[i].iLen1);
    x += chng.a[i].iLen1;
    blob_append_char(p->pOut, ',');
    blob_append_json_literal(p->pOut, 
                         pY->z + chng.a[i].iStart2, chng.a[i].iLen2);
  }
  blob_append_char(p->pOut, ',');
  blob_append_json_literal(p->pOut, pX->z + x, pX->n - x);
  blob_append(p->pOut, "],\n",3);
}
static void dfjsonEnd(DiffBuilder *p){
  blob_append(p->pOut, "0]", 2);
  fossil_free(p);
}
static DiffBuilder *dfjsonNew(Blob *pOut){
  DiffBuilder *p = fossil_malloc(sizeof(*p));
  p->xSkip = dfjsonSkip;
  p->xCommon = dfjsonCommon;
  p->xInsert = dfjsonInsert;
  p->xDelete = dfjsonDelete;
  p->xReplace = dfjsonReplace;
  p->xEdit = dfjsonEdit;
  p->xEnd = dfjsonEnd;
  p->lnLeft = p->lnRight = 0;
  p->pOut = pOut;
  blob_append_char(pOut, '[');
  return p;
}

/************************* DiffBuilderUnified********************************/
/* This formatter generates a unified diff for HTML.
**
** The result is a <table> with four columns.  The four columns hold:
**
**     1.   The line numbers for the first file.
**     2.   The line numbers for the second file.
**     3.   The "diff mark":  "+" or "-" or just a space
**     4.   Text of the line
**
** Inserted lines are marked with <ins> and deleted lines are marked
** with <del>.  The whole line is marked this way, not just the part that
** changed.  The part that change has an additional nested <ins> or <del>.
** The CSS needs to be set up such that a single <ins> or <del> gives a
** light background and a nested <ins> or <del> gives a darker background.
** Additional attributes (like bold font) might also be added to nested
** <ins> and <del> since those are the characters that have actually
** changed.
**
** Accumulator strategy:
**
**    *   Delete line numbers are output directly to p->pOut
**    *   Insert line numbers accumulate in p->aCol[0].
**    *   Separator marks accumulate in p->aCol[1].
**    *   Change text accumulates in p->aCol[2].
**    *   Pending insert line numbers go into p->aCol[3].
**    *   Pending insert text goes into p->aCol[4].
**
** eState is 1 if text has an open <del>
*/
static void dfunifiedFinishDelete(DiffBuilder *p){
  if( p->eState==0 ) return;
  blob_append(p->pOut, "</del>", 6);
  blob_append(&p->aCol[2], "</del>", 6);
  p->eState = 0;
}
static void dfunifiedFinishInsert(DiffBuilder *p){
  unsigned int i;
  if( p->nPending==0 ) return;
  dfunifiedFinishDelete(p);

  /* Blank lines for delete line numbers for each inserted line */
  for(i=0; i<p->nPending; i++) blob_append_char(p->pOut, '\n');

  /* Insert line numbers */
  blob_append(&p->aCol[0], "<ins>", 5);
  blob_append_xfer(&p->aCol[0], &p->aCol[3]);
  blob_append(&p->aCol[0], "</ins>", 6);

  /* "+" marks for the separator on inserted lines */
  for(i=0; i<p->nPending; i++) blob_append(&p->aCol[1], "+\n", 2);

  /* Text of the inserted lines */
  blob_append(&p->aCol[2], "<ins>", 5);
  blob_append_xfer(&p->aCol[2], &p->aCol[4]);
  blob_append(&p->aCol[2], "</ins>", 6);
  
  p->nPending = 0;
}
static void dfunifiedFinishRow(DiffBuilder *p){
  dfunifiedFinishDelete(p);
  dfunifiedFinishInsert(p);
  if( blob_size(&p->aCol[0])==0 ) return;
  blob_append(p->pOut, "</pre></td><td class=\"diffln difflnr\"><pre>\n", -1);
  blob_append_xfer(p->pOut, &p->aCol[0]);
  blob_append(p->pOut, "</pre></td><td class=\"diffsep\"><pre>\n", -1);
  blob_append_xfer(p->pOut, &p->aCol[1]);
  blob_append(p->pOut, "</pre></td><td class=\"difftxt difftxtu\"><pre>\n",-1);
  blob_append_xfer(p->pOut, &p->aCol[2]);
  blob_append(p->pOut, "</pre></td></tr>\n", -1);
}
static void dfunifiedStartRow(DiffBuilder *p){
  if( blob_size(&p->aCol[0])>0 ) return;
  blob_appendf(p->pOut,"<tr id=\"chunk%d\">"
                       "<td class=\"diffln difflnl\"><pre>\n", ++nChunk);
  p->eState = 0;
  p->nPending = 0;
}
static void dfunifiedSkip(DiffBuilder *p, unsigned int n, int isFinal){
  dfunifiedFinishRow(p);
  blob_append(p->pOut, "<tr><td class=\"diffln difflne\">"
                       "&#xfe19;</td><td></td><td></td></tr>\n", -1);
  p->lnLeft += n;
  p->lnRight += n;
}
static void dfunifiedCommon(DiffBuilder *p, const DLine *pLine){
  dfunifiedStartRow(p);
  dfunifiedFinishDelete(p);
  dfunifiedFinishInsert(p);
  p->lnLeft++;
  p->lnRight++;
  blob_appendf(p->pOut,"%d\n", p->lnLeft);
  blob_appendf(&p->aCol[0],"%d\n",p->lnRight);
  blob_append_char(&p->aCol[1], '\n');
  htmlize_to_blob(&p->aCol[2], pLine->z, (int)pLine->n);
  blob_append_char(&p->aCol[2], '\n');
}
static void dfunifiedInsert(DiffBuilder *p, const DLine *pLine){
  dfunifiedStartRow(p);
  p->lnRight++;
  blob_appendf(&p->aCol[3],"%d\n", p->lnRight);
  blob_append(&p->aCol[4], "<ins>", 5);
  htmlize_to_blob(&p->aCol[4], pLine->z, (int)pLine->n);
  blob_append(&p->aCol[4], "</ins>\n", 7);
  p->nPending++;
}
static void dfunifiedDelete(DiffBuilder *p, const DLine *pLine){
  dfunifiedStartRow(p);
  dfunifiedFinishInsert(p);
  if( p->eState==0 ){
    dfunifiedFinishInsert(p);
    blob_append(p->pOut, "<del>", 5);
    blob_append(&p->aCol[2], "<del>", 5);
    p->eState = 1;
  }
  p->lnLeft++;
  blob_appendf(p->pOut,"%d\n", p->lnLeft);
  blob_append_char(&p->aCol[0],'\n');
  blob_append(&p->aCol[1],"-\n",2);
  blob_append(&p->aCol[2], "<del>", 5);
  htmlize_to_blob(&p->aCol[2], pLine->z, (int)pLine->n);
  blob_append(&p->aCol[2], "</del>\n", 7);
}
static void dfunifiedReplace(DiffBuilder *p, const DLine *pX, const DLine *pY){
  dfunifiedStartRow(p);
  if( p->eState==0 ){
    dfunifiedFinishInsert(p);
    blob_append(p->pOut, "<del>", 5);
    blob_append(&p->aCol[2], "<del>", 5);
    p->eState = 1;
  }
  p->lnLeft++;
  p->lnRight++;
  blob_appendf(p->pOut,"%d\n", p->lnLeft);
  blob_append_char(&p->aCol[0], '\n');
  blob_append(&p->aCol[1], "-\n", 2);

  htmlize_to_blob(&p->aCol[2], pX->z,  pX->n);
  blob_append_char(&p->aCol[2], '\n');

  blob_appendf(&p->aCol[3],"%d\n", p->lnRight);

  htmlize_to_blob(&p->aCol[4], pY->z,  pY->n);
  blob_append_char(&p->aCol[4], '\n');
  p->nPending++;
}
static void dfunifiedEdit(DiffBuilder *p, const DLine *pX, const DLine *pY){
  int i;
  int x;
  LineChange chng;
  oneLineChange(pX, pY, &chng);
  dfunifiedStartRow(p);
  if( p->eState==0 ){
    dfunifiedFinishInsert(p);
    blob_append(p->pOut, "<del>", 5);
    blob_append(&p->aCol[2], "<del>", 5);
    p->eState = 1;
  }
  p->lnLeft++;
  p->lnRight++;
  blob_appendf(p->pOut,"%d\n", p->lnLeft);
  blob_append_char(&p->aCol[0], '\n');
  blob_append(&p->aCol[1], "-\n", 2);

  for(i=x=0; i<chng.n; i++){
    int ofst = chng.a[i].iStart1;
    int len = chng.a[i].iLen1;
    if( len ){
      htmlize_to_blob(&p->aCol[2], pX->z+x, ofst - x);
      x = ofst;
      blob_append(&p->aCol[2], "<del>", 5);
      htmlize_to_blob(&p->aCol[2], pX->z+x, len);
      x += len;
      blob_append(&p->aCol[2], "</del>", 6);
    }
  }
  htmlize_to_blob(&p->aCol[2], pX->z+x,  pX->n - x);
  blob_append_char(&p->aCol[2], '\n');

  blob_appendf(&p->aCol[3],"%d\n", p->lnRight);
  for(i=x=0; i<chng.n; i++){
    int ofst = chng.a[i].iStart2;
    int len = chng.a[i].iLen2;
    if( len ){
      htmlize_to_blob(&p->aCol[4], pY->z+x, ofst - x);
      x = ofst;
      blob_append(&p->aCol[4], "<ins>", 5);
      htmlize_to_blob(&p->aCol[4], pY->z+x, len);
      x += len;
      blob_append(&p->aCol[4], "</ins>", 6);
    }
  }
  htmlize_to_blob(&p->aCol[4], pY->z+x,  pY->n - x);
  blob_append_char(&p->aCol[4], '\n');
  p->nPending++;
}
static void dfunifiedEnd(DiffBuilder *p){
  dfunifiedFinishRow(p);
  blob_append(p->pOut, "</table>\n",-1);
  fossil_free(p);
}
static DiffBuilder *dfunifiedNew(Blob *pOut){
  DiffBuilder *p = fossil_malloc(sizeof(*p));
  p->xSkip = dfunifiedSkip;
  p->xCommon = dfunifiedCommon;
  p->xInsert = dfunifiedInsert;
  p->xDelete = dfunifiedDelete;
  p->xReplace = dfunifiedReplace;
  p->xEdit = dfunifiedEdit;
  p->xEnd = dfunifiedEnd;
  p->lnLeft = p->lnRight = 0;
  p->eState = 0;
  p->nPending = 0;
  p->pOut = pOut;
  blob_append(pOut, "<table class=\"diff udiff\">\n", -1);
  blob_init(&p->aCol[0], 0, 0);
  blob_init(&p->aCol[1], 0, 0);
  blob_init(&p->aCol[2], 0, 0);
  blob_init(&p->aCol[3], 0, 0);
  blob_init(&p->aCol[4], 0, 0);
  return p;
}

/************************* DiffBuilderSplit  ******************************/
/* This formatter creates a side-by-side diff in HTML.  The output is a
** <table> with 5 columns:
**
**    1.  Line numbers for the first file.
**    2.  Text for the first file.
**    3.  The difference mark.  "<", ">", "|" or blank
**    4.  Line numbers for the second file.
**    5.  Text for the second file.
**
** The <ins> and <del> strategy is the same as for unified diff above.
** In fact, the same CSS can be used for both.
**
** Accumulator strategy:
**
**    *   Left line numbers are output directly to p->pOut
**    *   Left text accumulates in p->aCol[0].
**    *   Edit marks accumulates in p->aCol[1].
**    *   Right line numbers accumulate in p->aCol[2].
**    *   Right text accumulates in p->aCol[3].
**
** eState:
**    0   In common block
**    1   Have <del> on the left
**    2   Have <ins> on the right
**    3   Have <del> on left and <ins> on the right
*/
static void dfsplitChangeState(DiffBuilder *p, int newState){
  if( p->eState == newState ) return;
  if( (p->eState&1)==0 && (newState & 1)!=0 ){
    blob_append(p->pOut, "<del>", 5);
    blob_append(&p->aCol[0], "<del>", 5);
    p->eState |= 1;
  }else if( (p->eState&1)!=0 && (newState & 1)==0 ){
    blob_append(p->pOut, "</del>", 6);
    blob_append(&p->aCol[0], "</del>", 6);
    p->eState &= ~1;
  }
  if( (p->eState&2)==0 && (newState & 2)!=0 ){
    blob_append(&p->aCol[2], "<ins>", 5);
    blob_append(&p->aCol[3], "<ins>", 5);
    p->eState |= 2;
  }else if( (p->eState&2)!=0 && (newState & 2)==0 ){
    blob_append(&p->aCol[2], "</ins>", 6);
    blob_append(&p->aCol[3], "</ins>", 6);
    p->eState &= ~2;
  }
}
static void dfsplitFinishRow(DiffBuilder *p){
  if( blob_size(&p->aCol[0])==0 ) return;
  dfsplitChangeState(p, 0);
  blob_append(p->pOut, "</pre></td><td class=\"difftxt difftxtl\"><pre>\n",-1);
  blob_append_xfer(p->pOut, &p->aCol[0]);
  blob_append(p->pOut, "</pre></td><td class=\"diffsep\"><pre>\n", -1);
  blob_append_xfer(p->pOut, &p->aCol[1]);
  blob_append(p->pOut, "</pre></td><td class=\"diffln difflnr\"><pre>\n",-1);
  blob_append_xfer(p->pOut, &p->aCol[2]);
  blob_append(p->pOut, "</pre></td><td class=\"difftxt difftxtr\"><pre>\n",-1);
  blob_append_xfer(p->pOut, &p->aCol[3]);
  blob_append(p->pOut, "</pre></td></tr>\n", -1);
}
static void dfsplitStartRow(DiffBuilder *p){
  if( blob_size(&p->aCol[0])>0 ) return;
  blob_appendf(p->pOut,"<tr id=\"chunk%d\">"
                       "<td class=\"diffln difflnl\"><pre>\n", ++nChunk);
  p->eState = 0;
}
static void dfsplitSkip(DiffBuilder *p, unsigned int n, int isFinal){
  dfsplitFinishRow(p);
  blob_append(p->pOut, 
     "<tr><td class=\"diffln difflnl difflne\">&#xfe19;</td>"
     "<td></td><td></td>"
     "<td class=\"diffln difflnr difflne\">&#xfe19;</td>"
     "<td/td></tr>\n", -1);
  p->lnLeft += n;
  p->lnRight += n;
}
static void dfsplitCommon(DiffBuilder *p, const DLine *pLine){
  dfsplitStartRow(p);
  dfsplitChangeState(p, 0);
  p->lnLeft++;
  p->lnRight++;
  blob_appendf(p->pOut,"%d\n", p->lnLeft);
  htmlize_to_blob(&p->aCol[0], pLine->z, (int)pLine->n);
  blob_append_char(&p->aCol[0], '\n');
  blob_append_char(&p->aCol[1], '\n');
  blob_appendf(&p->aCol[2],"%d\n",p->lnRight);
  htmlize_to_blob(&p->aCol[3], pLine->z, (int)pLine->n);
  blob_append_char(&p->aCol[3], '\n');
}
static void dfsplitInsert(DiffBuilder *p, const DLine *pLine){
  dfsplitStartRow(p);
  dfsplitChangeState(p, 2);
  p->lnRight++;
  blob_append_char(p->pOut, '\n');
  blob_append_char(&p->aCol[0], '\n');
  blob_append(&p->aCol[1], "&gt;\n", -1);
  blob_appendf(&p->aCol[2],"%d\n", p->lnRight);
  blob_append(&p->aCol[3], "<ins>", 5);
  htmlize_to_blob(&p->aCol[3], pLine->z, (int)pLine->n);
  blob_append(&p->aCol[3], "</ins>\n", 7);
}
static void dfsplitDelete(DiffBuilder *p, const DLine *pLine){
  dfsplitStartRow(p);
  dfsplitChangeState(p, 1);
  p->lnLeft++;
  blob_appendf(p->pOut,"%d\n", p->lnLeft);
  blob_append(&p->aCol[0], "<del>", 5);
  htmlize_to_blob(&p->aCol[0], pLine->z, (int)pLine->n);
  blob_append(&p->aCol[0], "</del>\n", 7);
  blob_append(&p->aCol[1], "&lt;\n", -1);
  blob_append_char(&p->aCol[2],'\n');
  blob_append_char(&p->aCol[3],'\n');
}
static void dfsplitReplace(DiffBuilder *p, const DLine *pX, const DLine *pY){
  dfsplitStartRow(p);
  dfsplitChangeState(p, 3);
  p->lnLeft++;
  p->lnRight++;
  blob_appendf(p->pOut,"%d\n", p->lnLeft);
  htmlize_to_blob(&p->aCol[0], pX->z,  pX->n);
  blob_append_char(&p->aCol[0], '\n');

  blob_append(&p->aCol[1], "|\n", 2);

  blob_appendf(&p->aCol[2],"%d\n", p->lnRight);

  htmlize_to_blob(&p->aCol[3], pY->z,  pY->n);
  blob_append_char(&p->aCol[3], '\n');
}
static void dfsplitEdit(DiffBuilder *p, const DLine *pX, const DLine *pY){
  int i;
  int x;
  LineChange chng;
  oneLineChange(pX, pY, &chng);
  dfsplitStartRow(p);
  dfsplitChangeState(p, 3);
  p->lnLeft++;
  p->lnRight++;
  blob_appendf(p->pOut,"%d\n", p->lnLeft);
  for(i=x=0; i<chng.n; i++){
    int ofst = chng.a[i].iStart1;
    int len = chng.a[i].iLen1;
    if( len ){
      htmlize_to_blob(&p->aCol[0], pX->z+x, ofst - x);
      x = ofst;
      if( chng.a[i].iLen2 ){
        blob_append(&p->aCol[0], "<del class='edit'>", -1);
      }else{
        blob_append(&p->aCol[0], "<del>", 5);
      }
      htmlize_to_blob(&p->aCol[0], pX->z+x, len);
      x += len;
      blob_append(&p->aCol[0], "</del>", 6);
    }
  }
  htmlize_to_blob(&p->aCol[0], pX->z+x,  pX->n - x);
  blob_append_char(&p->aCol[0], '\n');

  blob_append(&p->aCol[1], "|\n", 2);

  blob_appendf(&p->aCol[2],"%d\n", p->lnRight);
  for(i=x=0; i<chng.n; i++){
    int ofst = chng.a[i].iStart2;
    int len = chng.a[i].iLen2;
    if( len ){
      htmlize_to_blob(&p->aCol[3], pY->z+x, ofst - x);
      x = ofst;
      if( chng.a[i].iLen1 ){
        blob_append(&p->aCol[3], "<ins class='edit'>", -1);
      }else{
        blob_append(&p->aCol[3], "<ins>", 5);
      }
      htmlize_to_blob(&p->aCol[3], pY->z+x, len);
      x += len;
      blob_append(&p->aCol[3], "</ins>", 6);
    }
  }
  htmlize_to_blob(&p->aCol[3], pY->z+x,  pY->n - x);
  blob_append_char(&p->aCol[3], '\n');
}
static void dfsplitEnd(DiffBuilder *p){
  dfsplitFinishRow(p);
  blob_append(p->pOut, "</table>\n",-1);
  fossil_free(p);
}
static DiffBuilder *dfsplitNew(Blob *pOut){
  DiffBuilder *p = fossil_malloc(sizeof(*p));
  p->xSkip = dfsplitSkip;
  p->xCommon = dfsplitCommon;
  p->xInsert = dfsplitInsert;
  p->xDelete = dfsplitDelete;
  p->xReplace = dfsplitReplace;
  p->xEdit = dfsplitEdit;
  p->xEnd = dfsplitEnd;
  p->lnLeft = p->lnRight = 0;
  p->eState = 0;
  p->pOut = pOut;
  blob_append(pOut, "<table class=\"diff splitdiff\">\n", -1);
  blob_init(&p->aCol[0], 0, 0);
  blob_init(&p->aCol[1], 0, 0);
  blob_init(&p->aCol[2], 0, 0);
  blob_init(&p->aCol[3], 0, 0);
  blob_init(&p->aCol[4], 0, 0);
  return p;
}

/************************* DiffBuilderSbs  ******************************/
/* This formatter creates a side-by-side diff in text.
*/
static void dfsbsSkip(DiffBuilder *p, unsigned int n, int isFinal){
  if( (p->lnLeft || p->lnRight) && !isFinal ){
    blob_appendf(p->pOut, "%.*c\n", p->width*2 + 16, '.');
  }
  p->lnLeft += n;
  p->lnRight += n;
}

/*
** Append at least iMin characters (not bytes) and at most iMax characters
** from pX onto the into of p.
**
** This comment contains multibyte unicode characters (ü, Æ, ð) in order
** to test the ability of the diff code to handle such characters.
*/
static void sbs_append_chars(Blob *p, int iMin, int iMax, const DLine *pX){
  int i;
  const char *z = pX->z;
  for(i=0; i<iMax && i<pX->n; i++){
    char c = z[i];
    blob_append_char(p, c);
    if( (c&0xc0)==0x80 ){ iMin++; iMax++; }
  }
  while( i<iMin ){
    blob_append_char(p, ' ');
    i++;
  }
}

static void dfsbsCommon(DiffBuilder *p, const DLine *pLine){
  p->lnLeft++;
  p->lnRight++;
  blob_appendf(p->pOut,"%6u ", p->lnLeft);
  sbs_append_chars(p->pOut, p->width, p->width, pLine);
  blob_appendf(p->pOut,"   %6u ", p->lnRight);
  sbs_append_chars(p->pOut, 0, p->width, pLine);
  blob_append_char(p->pOut, '\n');
}
static void dfsbsInsert(DiffBuilder *p, const DLine *pLine){
  p->lnRight++;
  blob_appendf(p->pOut,"%6s %*s > %6u ",
     "", p->width, "", p->lnRight);
  sbs_append_chars(p->pOut, 0, p->width, pLine);
  blob_append_char(p->pOut, '\n');
}
static void dfsbsDelete(DiffBuilder *p, const DLine *pLine){
  p->lnLeft++;
  blob_appendf(p->pOut,"%6u ", p->lnLeft);
  sbs_append_chars(p->pOut, p->width, p->width, pLine);
  blob_append(p->pOut," <\n", 3);
}
static void dfsbsEdit(DiffBuilder *p, const DLine *pX, const DLine *pY){
  p->lnLeft++;
  p->lnRight++;
  blob_appendf(p->pOut,"%6u ", p->lnLeft);
  sbs_append_chars(p->pOut, p->width, p->width, pX);
  blob_appendf(p->pOut, " | %6u ", p->lnRight);
  sbs_append_chars(p->pOut, 0, p->width, pY);
  blob_append_char(p->pOut, '\n');
}
static void dfsbsEnd(DiffBuilder *p){
  fossil_free(p);
}
static DiffBuilder *dfsbsNew(Blob *pOut, u64 diffFlags){
  DiffBuilder *p = fossil_malloc(sizeof(*p));
  p->xSkip = dfsbsSkip;
  p->xCommon = dfsbsCommon;
  p->xInsert = dfsbsInsert;
  p->xDelete = dfsbsDelete;
  p->xReplace = dfsbsEdit;
  p->xEdit = dfsbsEdit;
  p->xEnd = dfsbsEnd;
  p->lnLeft = p->lnRight = 0;
  p->width = diff_width(diffFlags);
  p->pOut = pOut;
  return p;
}
/****************************************************************************/
/*
** Return the number between 0 and 100 that is smaller the closer pA and
** pB match.  Return 0 for a perfect match.  Return 100 if pA and pB are
** completely different.
**
** The current algorithm is as follows:
**
** (1) Remove leading and trailing whitespace.
** (2) Truncate both strings to at most 250 characters
** (3) Find the length of the longest common subsequence
** (4) Longer common subsequences yield lower scores.
*/
static int match_dline(const DLine *pA, const DLine *pB){
  const char *zA;            /* Left string */
  const char *zB;            /* right string */
  int nA;                    /* Bytes in zA[] */
  int nB;                    /* Bytes in zB[] */
  int avg;                   /* Average length of A and B */
  int i, j, k;               /* Loop counters */
  int best = 0;              /* Longest match found so far */

  fprintf(stderr, "A: [%.*s]\nB: [%.*s]\nbest=%d avg=%d score=%d\n",
  nA, zA+1, nB, zB+1, best, avg, score);
#endif

  /* Return the result */
  return score;
}

/*
** COMMAND:  test-line-match
** Usage: %fossil test-line-match STRING1 STRING2
**
** Return a score from 0 to 100 that is how similar STRING1 is to
** STRING2.  Smaller numbers mean more similar.  0 is an exact match.
**
** This command is used to test to match_dline() function in the
** internal Fossil diff logic.
*/
void test_dline_match(void){
  DLine a, b;
  int x;
  if( g.argc!=4 ) usage("STRING1 STRING2");
  a.z = g.argv[2];
  a.n = (int)strlen(a.z);
  b.z = g.argv[3];
  b.n = (int)strlen(b.z);
  x = match_dline(&a, &b);
  fossil_print("%d\n", x);
}

/*
** There is a change block in which nLeft lines of text on the left are
** converted into nRight lines of text on the right.  This routine computes
** how the lines on the left line up with the lines on the right.
**
** The return value is a buffer of unsigned characters, obtained from
** fossil_malloc().  (The caller needs to free the return value using
** fossil_free().)  Entries in the returned array have values as follows:
**
**    1.  Delete the next line of pLeft.
**    2.  Insert the next line of pRight.
**    3.  The next line of pLeft changes into the next line of pRight.
**    4.  Delete one line from pLeft and add one line to pRight.
**


** The length of the returned array will be at most nLeft+nRight bytes.
** If the first bytes is 4,  that means we could not compute reasonable
** alignment between the two blocks.
**
** Algorithm:  Wagner's minimum edit-distance algorithm, modified by
** adding a cost to each match based on how well the two rows match
** each other.  Insertion and deletion costs are 50.  Match costs
** are between 0 and 100 where 0 is a perfect match 100 is a complete
** mismatch.
*/
static unsigned char *diffBlockAlignment(
  const DLine *aLeft, int nLeft,     /* Text on the left */
  const DLine *aRight, int nRight,   /* Text on the right */
  u64 diffFlags,                     /* Flags passed into the original diff */
  int *pNResult                      /* OUTPUT: Bytes of result */
){
  int i, j, k;                 /* Loop counters */
  int *a;                      /* One row of the Wagner matrix */
  int *pToFree;                /* Space that needs to be freed */
  unsigned char *aM;           /* Wagner result matrix */
  int nMatch, iMatch;          /* Number of matching lines and match score */
  int mnLen;                   /* minInt(nLeft, nRight) */
  int mxLen;                   /* MAX(nLeft, nRight) */
  int aBuf[100];               /* Stack space for a[] if nRight not to big */


  if( nLeft==0 ){
    aM = fossil_malloc( nRight + 2 );
    memset(aM, 2, nRight);
    *pNResult = nRight;
    return aM;
  }
  if( nRight==0 ){
    aM = fossil_malloc( nLeft + 2 );
    memset(aM, 1, nLeft);
    *pNResult = nLeft;
    return aM;
  }

  /* For large alignments, use a divide and conquer algorithm that is
  ** O(NlogN).  The result is not as precise, but this whole thing is an
  ** approximation anyhow, and the faster response time is an acceptable
  ** trade-off for reduced precision.
  */
  mnLen = nLeft<nRight ? nLeft : nRight;
  if( nLeft*nRight>1000 && (diffFlags & DIFF_SLOW_SBS)==0 ){
    const DLine *aSmall;   /* The smaller of aLeft and aRight */
    const DLine *aBig;     /* The larger of aLeft and aRight */
    int nSmall, nBig;      /* Size of aSmall and aBig.  nSmall<=nBig */
    int iDivSmall, iDivBig;  /* Divider point for aSmall and aBig */
    int iDivLeft, iDivRight; /* Divider point for aLeft and aRight */
    unsigned char *a1, *a2;  /* Results of the alignments on two halves */
    int n1, n2;              /* Number of entries in a1 and a2 */
    int score, bestScore;    /* Score and best score seen so far */
    if( nLeft>nRight ){
      aSmall = aRight;
      nSmall = nRight;
      aBig = aLeft;
      nBig = nLeft;
    }else{
      aSmall = aLeft;
      nSmall = nLeft;
      aBig = aRight;
      nBig = nRight;
    }


    iDivBig = nBig/2;
    iDivSmall = nSmall/2;
    bestScore = 10000;
    for(i=0; i<nSmall; i++){
      score = match_dline(aBig+iDivBig, aSmall+i) + abs(i-nSmall/2)*2;
      if( score<bestScore ){
        bestScore = score;
        iDivSmall = i;
      }
    }
    if( aSmall==aRight ){
      iDivRight = iDivSmall;
      iDivLeft = iDivBig;
    }else{
      iDivRight = iDivBig;
      iDivLeft = iDivSmall;
    }
    a1 = diffBlockAlignment(aLeft,iDivLeft,aRight,iDivRight,diffFlags,&n1);
    a2 = diffBlockAlignment(aLeft+iDivLeft, nLeft-iDivLeft,
                            aRight+iDivRight, nRight-iDivRight,
                            diffFlags, &n2);
    a1 = fossil_realloc(a1, n1+n2 );
    memcpy(a1+n1,a2,n2);
    fossil_free(a2);
    *pNResult = n1+n2;
    return a1;
  }

  /* If we reach this point, we will be doing an O(N*N) Wagner minimum
  ** edit distance to compute the alignment.
  */
  if( nRight < count(aBuf)-1 ){
    pToFree = 0;
    a = aBuf;
  }else{
    a = pToFree = fossil_malloc( sizeof(a[0])*(nRight+1) );
  }
  aM = fossil_malloc( (nLeft+1)*(nRight+1) );


  /* Compute the best alignment */
  for(i=0; i<=nRight; i++){
    aM[i] = 2;
    a[i] = i*50;
  }
  aM[0] = 0;

    }
    k--;
    aM[k] = aM[j*(nRight+1)+i];
  }
  k++;
  i = (nRight+1)*(nLeft+1) - k;
  memmove(aM, &aM[k], i);
  *pNResult = i;

  /* If:
  **   (1) the alignment is more than 25% longer than the longest side, and
  **   (2) the average match cost exceeds 15
  ** Then this is probably an alignment that will be difficult for humans
  ** to read.  So instead, just show all of the right side inserted followed
  ** by all of the left side deleted.
  **
  ** The coefficients for conditions (1) and (2) above are determined by
  ** experimentation.
  */
  mxLen = nLeft>nRight ? nLeft : nRight;
  if( i*4>mxLen*5 && (nMatch==0 || iMatch/nMatch>15) ){
    memset(aM, 4, mnLen); *pNResult = mnLen;
    if( nLeft>mnLen ){  memset(aM+mnLen, 1, nLeft-mnLen); *pNResult = nLeft;   }
    if( nRight>mnLen ){ memset(aM+mnLen, 2, nRight-mnLen); *pNResult = nRight; }
  }

  /* Return the result */
  fossil_free(pToFree);
  return aM;
}

/*
** R[] is an array of six integer, two COPY/DELETE/INSERT triples for a
** pair of adjacent differences.  Return true if the gap between these
** two differences is so small that they should be rendered as a single
** edit.
*/
static int smallGap(const int *R){
  return R[3]<=2 || R[3]<=(R[1]+R[2]+R[4]+R[5])/8;
}

/*

** Format a diff using a DiffBuilder object
*/
static void formatDiff(
  DContext *p,           /* The computed diff */

  ReCompiled *pRe,       /* Only show changes that match this regex */
  u64 diffFlags,         /* Flags controlling the diff */
  DiffBuilder *pBuilder  /* The formatter object */
){
  const DLine *A;        /* Left side of the diff */
  const DLine *B;        /* Right side of the diff */
  unsigned int a = 0;    /* Index of next line in A[] */
  unsigned int b = 0;    /* Index of next line in B[] */
  const int *R;          /* Array of COPY/DELETE/INSERT triples */
  unsigned int r;        /* Index into R[] */
  unsigned int nr;       /* Number of COPY/DELETE/INSERT triples to process */
  unsigned int mxr;      /* Maximum value for r */
  unsigned int na, nb;   /* Number of lines shown from A and B */
  unsigned int i, j;     /* Loop counters */
  unsigned int m, ma, mb;/* Number of lines to output */
  signed int skip = 0;   /* Number of lines to skip */


  unsigned int nContext; /* Lines of context above and below each change */





  nContext = diff_context_lines(diffFlags);















  A = p->aFrom;
  B = p->aTo;
  R = p->aEdit;
  mxr = p->nEdit;
  while( mxr>2 && R[mxr-1]==0 && R[mxr-2]==0 ){ mxr -= 3; }

  for(r=0; r<mxr; r += 3*nr){
    /* Figure out how many triples to show in a single block */
    for(nr=1; R[r+nr*3]>0 && R[r+nr*3]<nContext*2; nr++){}


    /* If there is a regex, skip this block (generate no diff output)
    ** if the regex matches or does not match both insert and delete.
    ** Only display the block if one side matches but the other side does
    ** not.
    */
    if( pRe ){

      if( hideBlock ){
        a = xa;
        b = xb;
        continue;
      }
    }


    /* Figure out how many lines of A and B are to be displayed
    ** for this change block.
    */
    if( R[r]>nContext ){
      na = nb = nContext;
      skip = R[r] - nContext;
    }else{
      na = nb = R[r];
      skip = 0;

      nb += R[r+nr*3];
    }
    for(i=1; i<nr; i++){
      na += R[r+i*3];
      nb += R[r+i*3];
    }






















    /* Show the initial common area */
    a += skip;
    b += skip;
    m = R[r] - skip;
    if( r ) skip -= nContext;
    if( skip>0 ){
      pBuilder->xSkip(pBuilder, skip, 0);
    }
    for(j=0; j<m; j++){


      pBuilder->xCommon(pBuilder, &A[a+j]);



    }
    a += m;
    b += m;

    /* Show the differences */
    for(i=0; i<nr; i++){
      int nAlign;
      unsigned char *alignment;
      ma = R[r+i*3+1];   /* Lines on left but not on right */
      mb = R[r+i*3+2];   /* Lines on right but not on left */

      /* Try to find an alignment for the lines within this one block */
      alignment = diffBlockAlignment(&A[a], ma, &B[b], mb, diffFlags, &nAlign);

      /* If we could not get a good alignment, try merging the current
      ** block with subsequent blocks, if the subsequent blocks are
      ** nearby */
      while( alignment[0]==4 && i<nr-1 && smallGap(&R[r+i*3]) ){
        i++;
        m = R[r+i*3];
        ma += R[r+i*3+1] + m;
        mb += R[r+i*3+2] + m;
        fossil_free(alignment);
        alignment = diffBlockAlignment(&A[a], ma, &B[b], mb, diffFlags,&nAlign);
      }


      for(j=0; ma+mb>0; j++){
        assert( j<nAlign );
        switch( alignment[j] ){
          case 1: {
            /* Delete one line from the left */



            pBuilder->xDelete(pBuilder, &A[a]);
            ma--;
            a++;
            break;
          }
          case 2: {
            /* Insert one line on the right */
            pBuilder->xInsert(pBuilder, &B[b]);
            assert( mb>0 );
            mb--;
            b++;
            break;
          }
          case 3: {
            /* The left line is changed into the right line */
            pBuilder->xEdit(pBuilder, &A[a], &B[b]);
            assert( ma>0 && mb>0 );
            ma--;
            mb--;
            a++;
            b++;
            break;



          }









          case 4: {
            /* Delete from left then separately insert on the right */




            pBuilder->xReplace(pBuilder, &A[a], &B[b]);






            ma--;

            a++;
            mb--;
            b++;
            break;
          }
        }
      }
      assert( nAlign==j );
      fossil_free(alignment);
      if( i<nr-1 ){
        m = R[r+i*3+3];
        for(j=0; j<m; j++){


          pBuilder->xCommon(pBuilder, &A[a+j]);



        }
        b += m;
        a += m;
      }
    }

    /* Show the final common area */
    assert( nr==i );
    m = R[r+nr*3];
    if( m>nContext ) m = nContext;
    for(j=0; j<m && j<nContext; j++){


      pBuilder->xCommon(pBuilder, &A[a+j]);



    }
  }
  if( R[r]>nContext ){





    pBuilder->xSkip(pBuilder, R[r] - nContext, 1);
  }
  pBuilder->xEnd(pBuilder);
}


/*
** Compute the optimal longest common subsequence (LCS) using an
** exhaustive search.  This version of the LCS is only used for
** shorter input strings since runtime is O(N*N) where N is the
** input string length.
*/

){
  int i, j, k;               /* Loop counters */
  int n;                     /* Loop limit */
  DLine *pA, *pB;            /* Pointers to lines */
  int iSX, iSY, iEX, iEY;    /* Current match */
  int skew = 0;              /* How lopsided is the match */
  int dist = 0;              /* Distance of match from center */
  int mid;                   /* Center of the chng */
  int iSXb, iSYb, iEXb, iEYb;   /* Best match so far */
  int iSXp, iSYp, iEXp, iEYp;   /* Previous match */
  sqlite3_int64 bestScore;      /* Best score so far */
  sqlite3_int64 score;          /* Score for current candidate LCS */
  int span;                     /* combined width of the input sequences */

  span = (iE1 - iS1) + (iE2 - iS2);

    blob_appendf(pOut, "<p class=\"generalError\">%s</p>", msg);
  }else{
    blob_append(pOut, msg, -1);
  }
}

/*
** Generate a report of the differences between files pA_Blob and pB_Blob.
**
** If pOut!=NULL then append text to pOut that will be the difference,
** formatted according to flags in diffFlags.  The pOut Blob must have
** already been initialized.
**
** If pOut==NULL then no formatting occurs.  Instead, this routine
** returns a pointer to an array of integers.  The integers come in
** triples.  The elements of each triple are:
**
**   1.  The number of lines to copy
**   2.  The number of lines to delete
**   3.  The number of lines to insert
**
** The return vector is terminated bin a triple of all zeros.  The caller
** should free the returned vector using fossil_free().
**
** This diff utility does not work on binary files.  If a binary
** file is encountered, 0 is returned and pOut is written with
** text "cannot compute difference between binary files".
*/
int *text_diff(
  Blob *pA_Blob,   /* FROM file */

      }
      g.diffCnt[1] += nIns;
      g.diffCnt[2] += nDel;
      if( nIns+nDel ){
        g.diffCnt[0]++;
        blob_appendf(pOut, "%10d %10d", nIns, nDel);
      }
    }else if( diffFlags & DIFF_RAW ){
      const int *R = c.aEdit;
      unsigned int r;
      for(r=0; R[r] || R[r+1] || R[r+2]; r += 3){
        blob_appendf(pOut, " copy %6d  delete %6d  insert %6d\n",
                     R[r], R[r+1], R[r+2]);
      }
    }else if( diffFlags & DIFF_JSON ){
      DiffBuilder *pBuilder = dfjsonNew(pOut);
      formatDiff(&c, pRe, diffFlags, pBuilder);
      blob_append_char(pOut, '\n');
    }else if( diffFlags & DIFF_TCL ){
      DiffBuilder *pBuilder = dftclNew(pOut);
      formatDiff(&c, pRe, diffFlags, pBuilder);
    }else if( diffFlags & DIFF_SIDEBYSIDE ){
      DiffBuilder *pBuilder;
      if( diffFlags & DIFF_HTML ){
        pBuilder = dfsplitNew(pOut);
      }else{
        pBuilder = dfsbsNew(pOut, diffFlags);
      }
      formatDiff(&c, pRe, diffFlags, pBuilder);
    }else if( diffFlags & DIFF_DEBUG ){
      DiffBuilder *pBuilder = dfdebugNew(pOut);
      formatDiff(&c, pRe, diffFlags, pBuilder);
    }else if( diffFlags & DIFF_HTML ){
      DiffBuilder *pBuilder = dfunifiedNew(pOut);
      formatDiff(&c, pRe, diffFlags, pBuilder);
    }else{
      contextDiff(&c, pOut, diffFlags);
    }
    fossil_free(c.aFrom);
    fossil_free(c.aTo);
    fossil_free(c.aEdit);
    return 0;
  }else{
    /* If a context diff is not requested, then return the

  if( find_option("browser","b",0)!=0 ){
    diffFlags |= DIFF_HTML|DIFF_WEBPAGE|DIFF_LINENO|DIFF_BROWSER;
  }
  if( find_option("by",0,0)!=0 ){
    diffFlags |= DIFF_HTML|DIFF_WEBPAGE|DIFF_LINENO|DIFF_BROWSER
                   |DIFF_SIDEBYSIDE;
  }
  if( find_option("json",0,0)!=0 ){
    diffFlags |= DIFF_JSON;
  }
  if( find_option("tcl",0,0)!=0 ){














    diffFlags |= DIFF_TCL;








  }



  /* Undocumented and unsupported flags used for development
  ** debugging and analysis: */
  if( find_option("debug",0,0)!=0 ) diffFlags |= DIFF_DEBUG;
  if( find_option("raw",0,0)!=0 )   diffFlags |= DIFF_RAW;
  return diffFlags;
}

/*
** COMMAND: test-diff
** COMMAND: xdiff
**
** Usage: %fossil xdiff [options] FILE1 FILE2
**
** This is the "external diff" feature.  By "external" here we mean a diff
** applied to files that are not under version control.  See the "diff"
** command for computing differences between files that are under control.
**
** This command prints the differences between the two files FILE1 and FILE2.
** all of the usual diff command-line options apply.  See the "diff" command
** for a full list of command-line options.
**
** This command used to be called "test-diff".  The older "test-diff" spelling
** still works, for compatibility.
*/
void test_diff_cmd(void){
  Blob a, b, out;
  u64 diffFlag;
  const char *zRe;           /* Regex filter for diff output */
  ReCompiled *pRe = 0;       /* Regex filter for diff output */

/* Refinements to the display of unified and side-by-side diffs.
**
** In all cases, the table columns tagged with "difftxt" are expanded,
** where possible, to fill the width of the screen.
**
** For a side-by-side diff, if either column is two wide to fit on the
** display, scrollbars are added.  The scrollbars are linked, so that
** both sides scroll together.  Left and right arrows also scroll.
*/
(function(){
  var SCROLL_LEN = 25;
  function initDiff(diff){
    var txtCols = diff.querySelectorAll('td.difftxt');
    var txtPres = diff.querySelectorAll('td.difftxt pre');
    var width = 0;
    if(txtPres.length>=2)Math.max(txtPres[0].scrollWidth, txtPres[1].scrollWidth);
    var i;
    for(i=0; i<txtCols.length; i++){
      txtCols[i].style.width = width + 'px';
      txtPres[i].style.maxWidth = width + 'px';
      txtPres[i].style.width = width + 'px';
      txtPres[i].onscroll = function(e){
        for(var j=0; j<txtPres.length; j++) txtPres[j].scrollLeft = this.scrollLeft;
      };
    }
    diff.tabIndex = 0;
    diff.onkeydown = function(e){
      e = e || event;
      var len = {37: -SCROLL_LEN, 39: SCROLL_LEN}[e.keyCode];
      if( !len ) return;
      txtCols[0].scrollLeft += len;
      return false;
    };
  }
  var i, diffs = document.querySelectorAll('table.splitdiff')
  for(i=0; i<diffs.length; i++){
    initDiff(diffs[i]);
  }
  if(window.fossil && fossil.page){
    fossil.page.tweakSbsDiffs = function(){
      document.querySelectorAll('table.splitdiff').forEach(initDiff);
    };
  }
  var lastWidth = 0;
  function checkWidth(){
    if( document.body.clientWidth!=lastWidth ){
      lastWidth = document.body.clientWidth;
      var w = lastWidth*0.5 - 100;
      var allCols = document.querySelectorAll('td.difftxtl pre');
      for(let i=0; i<allCols.length; i++){
        allCols[i].style.width = w + "px";
        allCols[i].style.maxWidth = w + "px";
      }
      allCols = document.querySelectorAll('td.difftxtr pre');
      for(let i=0; i<allCols.length; i++){
        allCols[i].style.width = w + "px";
        allCols[i].style.maxWidth = w + "px";
      }
      var allDiffs = document.querySelectorAll('table.diff');
      w = lastWidth;
      for(let i=0; i<allDiffs.length; i++){
        allDiffs[i].style.width = '98%'; // setting to w causes unsightly horiz. scrollbar
        allDiffs[i].style.maxWidth = w + "px";
      }
    }
    setTimeout(checkWidth, 100)
  }
  checkWidth();
})();

  TXT_COL_BG #ffffff
  TXT_COL_FG #000000
  MKR_COL_BG #444444
  MKR_COL_FG #dddddd
  CHNG_BG    #d0d0ff
  ADD_BG     #c0ffc0
  RM_BG      #ffc0c0
  HR_FG      #444444
  HR_PAD_TOP 4
  HR_PAD_BTM 8
  FN_BG      #444444
  FN_FG      #ffffff
  FN_PAD     5
  ERR_FG     #ee0000
  PADX       5

    fconfigure $in -encoding utf-8
    set difftxt [split [read $in] \n]
    close $in
  }
  set N [llength $difftxt]
  set ii 0
  set nDiffs 0
  set n1 0
  set n2 0  
  array set widths {txt 0 ln 0 mkr 1}
  while {[set line [getLine $difftxt $N ii]] != -1} {


    switch -- [lindex $line 0] {
      FILE {
        incr nDiffs
        foreach wx [list [string length $n1] [string length $n2]] {
          if {$wx>$widths(ln)} {set widths(ln) $wx}
        }
        .lnA insert end \n fn \n -
        .txtA insert end [lindex $line 1]\n fn \n -
        .mkr insert end \n fn \n -
        .lnB insert end \n fn \n -
        .txtB insert end [lindex $line 2]\n fn \n -
        .wfiles.lb insert end [lindex $line 2]
        set n1 0
        set n2 0
      }
      SKIP {
        set n [lindex $line 1]
        incr n1 $n

        incr n2 $n
        .lnA insert end ...\n hrln
        .txtA insert end [string repeat . 30]\n hrtxt


        .mkr insert end \n hrln
        .lnB insert end ...\n hrln
        .txtB insert end [string repeat . 30]\n hrtxt
      }
      COM {
        set x [lindex $line 1]
        incr n1
        incr n2
        .lnA insert end $n1\n -

        .txtA insert end $x\n -
        .mkr insert end \n -
        .lnB insert end $n2\n -
        .txtB insert end $x\n -
      }
      INS {
        set x [lindex $line 1]
        incr n2
        .lnA insert end \n -
        .txtA insert end \n -
        .mkr insert end >\n -
        .lnB insert end $n2\n -
        .txtB insert end $x add \n -
      }
      DEL {

        set x [lindex $line 1]
        incr n1
        .lnA insert end $n1\n -
        .txtA insert end $x rm \n -
        .mkr insert end <\n -
        .lnB insert end \n -
        .txtB insert end \n -
      }
      EDIT {
        incr n1
        incr n2
        .lnA insert end $n1\n -
        .lnB insert end $n2\n -
        .mkr insert end |\n -
        set nn [llength $line]
        for {set i 1} {$i<$nn} {incr i 3} {
          set x [lindex $line $i]
          if {$x ne ""} {
            .txtA insert end $x -
            .txtB insert end $x -
          }
          if {$i+2<$nn} {
            set x1 [lindex $line [expr {$i+1}]]
            set x2 [lindex $line [expr {$i+2}]]
            if {"$x1" eq ""} {
              .txtB insert end $x2 add


            } elseif {"$x2" eq ""} {
              .txtA insert end $x1 rm
            } else {
              .txtA insert end $x1 chng
              .txtB insert end $x2 chng
            }
          }
        }
        .txtA insert end \n -
        .txtB insert end \n -
      }
      "" {

        foreach wx [list [string length $n1] [string length $n2]] {
          if {$wx>$widths(ln)} {set widths(ln) $wx}
        }
      }
      default {
        error "bad diff source line: $line"
      }
    }
  }

  foreach c [cols] {
    set type [colType $c]
    if {$type ne "txt"} {
      $c config -width $widths($type)

text .mkr

foreach c [cols] {
  set keyPrefix [string toupper [colType $c]]_COL_
  if {[tk windowingsystem] eq "win32"} {$c config -font {courier 9}}
  $c config -bg $CFG(${keyPrefix}BG) -fg $CFG(${keyPrefix}FG) -borderwidth 0 \
    -padx $CFG(PADX) -yscroll sync-y
  $c tag config hrln -spacing1 $CFG(HR_PAD_TOP) -spacing3 $CFG(HR_PAD_BTM) \
     -foreground $CFG(HR_FG) -justify right
  $c tag config hrtxt  -spacing1 $CFG(HR_PAD_TOP) -spacing3 $CFG(HR_PAD_BTM) \
     -foreground $CFG(HR_FG) -justify center
  $c tag config fn -spacing1 $CFG(FN_PAD) -spacing3 $CFG(FN_PAD)
  bindtags $c ". $c Text all"
  bind $c <1> {focus %W}
}

::ttk::scrollbar .sby -command {.txtA yview} -orient vertical
::ttk::scrollbar .sbxA -command {.txtA xview} -orient horizontal

  return 0;
}

/*
** Print the "Index:" message that patches wants to see at the top of a diff.
*/
void diff_print_index(const char *zFile, u64 diffFlags, Blob *diffBlob){
  if( (diffFlags & (DIFF_SIDEBYSIDE|DIFF_BRIEF|DIFF_NUMSTAT|DIFF_JSON|
                    DIFF_WEBPAGE|DIFF_TCL))==0 ){
    char *z = mprintf("Index: %s\n%.66c\n", zFile, '=');
    if( !diffBlob ){
      fossil_print("%s", z);
    }else{
      blob_appendf(diffBlob, "%s", z);
    }
    fossil_free(z);
  }
}

/*
** Print the +++/--- filename lines for a diff operation.
*/
void diff_print_filenames(
  const char *zLeft,
  const char *zRight,
  u64 diffFlags,
  Blob *diffBlob
){
  char *z = 0;
  if( diffFlags & (DIFF_BRIEF|DIFF_RAW|DIFF_JSON) ){
    /* no-op */
  }else if( diffFlags & DIFF_DEBUG ){
    fossil_print("FILE-LEFT   %s\nFILE-RIGHT  %s\n",
       zLeft, zRight);
  }else if( diffFlags & DIFF_WEBPAGE ){
    if( fossil_strcmp(zLeft,zRight)==0 ){
      z = mprintf("<h1>%h</h1>\n", zLeft);
    }else{
      z = mprintf("<h1>%h &lrarr; %h</h1>\n", zLeft, zRight);
    }
  }else if( diffFlags & DIFF_TCL ){
    Blob *pOut;
    Blob x;
    if( diffBlob ){
      pOut = diffBlob;
    }else{
      blob_init(&x, 0, 0);
      pOut = &x;
    }
    blob_append(pOut, "FILE ", 5);
    blob_append_tcl_literal(pOut, zLeft, (int)strlen(zLeft));
    blob_append_char(pOut, ' ');
    blob_append_tcl_literal(pOut, zRight, (int)strlen(zRight));
    blob_append_char(pOut, '\n');
    if( !diffBlob ){
      fossil_print("%s", blob_str(pOut));
      blob_reset(&x);
    }
    return;
  }else if( diffFlags & DIFF_SIDEBYSIDE ){
    int w = diff_width(diffFlags);
    int n1 = strlen(zLeft);
    int n2 = strlen(zRight);
    int x;
    if( n1==n2 && fossil_strcmp(zLeft,zRight)==0 ){
      if( n1>w*2 ) n1 = w*2;

    fossil_print("%s", z);
  }else{
    blob_appendf(diffBlob, "%s", z);
  }
  fossil_free(z);
}

































/*
** Default header text for diff with --webpage
*/
static const char zWebpageHdr[] = 
@ <!DOCTYPE html>
@ <html>
@ <head>
@ <meta charset="UTF-8">
@ <style>

















@ h1 {
@   font-size: 150%;
@ }
@
@ table.diff {
@   width: 98%;
@   border-spacing: 0;
@   border: 1px solid black;
@ }
@ table.diff td {
@   vertical-align: top;
@ }
@ table.diff pre {
@   margin: 0 0 0 0;
@ }
@ td.diffln {
@   width: 1px;
@   text-align: right;
@   padding: 0 1em 0 0;
@ }
@ td.difflne {
@   padding-bottom: 0.4em;
@ }
@ td.diffsep {
@   width: 1px;
@   padding: 0 0.3em 0 1em;
@ }
@ td.difftxt pre {
@   overflow-x: auto;
@ }
@ td.diffln ins {
@   background-color: #a0e4b2;
@   text-decoration: none;
@ }
@ td.diffln del {
@   background-color: #ffc0c0;
@   text-decoration: none;
@ }
@ td.difftxt del {
@   background-color: #ffe8e8;
@   text-decoration: none;
@ }
@ td.difftxt del > del {
@   background-color: #ffc0c0;
@   text-decoration: none;
@   font-weight: bold;
@ }
@ td.difftxt del > del.edit {
@   background-color: #c0c0ff;
@   text-decoration: none;
@   font-weight: bold;
@ }
@ td.difftxt ins {
@   background-color: #dafbe1;
@   text-decoration: none;
@ }
@ td.difftxt ins > ins {
@   background-color: #a0e4b2;
@   text-decoration: none;
@   font-weight: bold;
@ }
@ td.difftxt ins > ins.edit {
@   background-color: #c0c0ff;
@   text-decoration: none;
@   font-weight: bold;
@ }
@ 
@ </style>
@ </head>
@ <body>
;
const char zWebpageEnd[] = 
@ </body>
@ </html>

#ifndef _WIN32
    signal(SIGINT, diff_www_interrupt);
#else
    SetConsoleCtrlHandler(diff_console_ctrl_handler, TRUE);
#endif
  }
  if( (diffFlags & DIFF_WEBPAGE)!=0 ){






    fossil_print("%s",zWebpageHdr);
    fflush(stdout);
  }
}

/* Do any final output required by a diff and complete the diff
** process.
**
** For --browser and --webpage, output any javascript required by 
** the diff.  (Currently JS is only needed for side-by-side diffs).
**
** For --browser, close the connection to the temporary file, then
** launch a web browser to view the file.  After a delay
** of FOSSIL_BROWSER_DIFF_DELAY milliseconds, delete the temp file.
*/
void diff_end(u64 diffFlags, int nErr){
  if( (diffFlags & DIFF_WEBPAGE)!=0 ){
    if( diffFlags & DIFF_SIDEBYSIDE ){
      const unsigned char *zJs = builtin_file("diff.js", 0);
      fossil_print("<script>\n%s</script>\n", zJs);
    }
    fossil_print("%s", zWebpageEnd);
  }
  if( (diffFlags & DIFF_BROWSER)!=0 && nErr==0 ){
    char *zCmd = mprintf("%s %$", fossil_web_browser(), tempDiffFilename);
    fclose(diffOut);

){
  int vid;
  Blob sql;
  Stmt q;
  int asNewFile;            /* Treat non-existant files as empty files */
  int isNumStat;            /* True for --numstat */

  asNewFile = (diffFlags & (DIFF_VERBOSE|DIFF_NUMSTAT|DIFF_HTML))!=0;
  isNumStat = (diffFlags & (DIFF_NUMSTAT|DIFF_TCL|DIFF_HTML))!=0;
  vid = db_lget_int("checkout", 0);
  vfile_check_signature(vid, CKSIG_ENOTFILE);
  blob_zero(&sql);
  db_begin_transaction();
  if( zFrom ){
    int rid = name_to_typed_rid(zFrom, "ci");
    if( !is_a_version(rid) ){

    }else if( pToFile==0 ){
      cmp = -1;
    }else{
      cmp = fossil_strcmp(pFromFile->zName, pToFile->zName);
    }
    if( cmp<0 ){
      if( file_dir_match(pFileDir, pFromFile->zName) ){
        if( (diffFlags & (DIFF_NUMSTAT|DIFF_HTML))==0 ){
          fossil_print("DELETED %s\n", pFromFile->zName);
        }
        if( asNewFlag ){
          diff_manifest_entry(pFromFile, 0, zDiffCmd, zBinGlob,
                              fIncludeBinary, diffFlags);
        }
      }
      pFromFile = manifest_file_next(pFrom,0);
    }else if( cmp>0 ){
      if( file_dir_match(pFileDir, pToFile->zName) ){
        if( (diffFlags & (DIFF_NUMSTAT|DIFF_HTML|DIFF_TCL|DIFF_JSON))==0 ){
          fossil_print("ADDED   %s\n", pToFile->zName);
        }
        if( asNewFlag ){
          diff_manifest_entry(0, pToFile, zDiffCmd, zBinGlob,
                              fIncludeBinary, diffFlags);
        }
      }

void diff_tk(const char *zSubCmd, int firstArg){
  int i;
  Blob script;
  const char *zTempFile = 0;
  char *zCmd;
  const char *zTclsh;
  blob_zero(&script);
  blob_appendf(&script, "set fossilcmd {| \"%/\" %s -tcl -i -v",
               g.nameOfExe, zSubCmd);
  find_option("tcl",0,0);
  find_option("html",0,0);
  find_option("side-by-side","y",0);
  find_option("internal","i",0);
  find_option("verbose","v",0);
  zTclsh = find_option("tclsh",0,1);
  if( zTclsh==0 ){
    zTclsh = db_get("tclsh",0);

  ** used for dynamically toggling certain UI components on and off.
  ** Must come after window.fossil has been intialized and before
  ** fossil.page.fileedit.js. Potential TODO: move this into the
  ** window.fossil bootstrapping so that we don't have to "fulfill"
  ** the JS multiple times.
  */
  ajax_emit_js_preview_modes(1);
  builtin_request_js("diff.js");
  builtin_request_js("fossil.page.fileedit.js");
  builtin_fulfill_js_requests();
  {
    /* Dynamically populate the editor, display any error in the err
    ** blob, and/or switch to tab #0, where the file selector
    ** lives. The extra C scopes here correspond to JS-level scopes,
    ** to improve grokability. */

}

/*
** Generate javascript to enhance HTML diffs.
*/
void append_diff_javascript(int sideBySide){
  if( !sideBySide ) return;
  builtin_request_js("diff.js");
}

/*
** Construct an appropriate diffFlag for text_diff() based on query
** parameters and the to boolean arguments.
*/
u64 construct_diff_flags(int diffType){

  $(SRCDIR)/alerts/bflat3.wav \
  $(SRCDIR)/alerts/bloop.wav \
  $(SRCDIR)/alerts/plunk.wav \
  $(SRCDIR)/chat.js \
  $(SRCDIR)/ci_edit.js \
  $(SRCDIR)/copybtn.js \
  $(SRCDIR)/default.css \
  $(SRCDIR)/diff.js \
  $(SRCDIR)/diff.tcl \
  $(SRCDIR)/forum.js \
  $(SRCDIR)/fossil.bootstrap.js \
  $(SRCDIR)/fossil.confirmer.js \
  $(SRCDIR)/fossil.copybutton.js \
  $(SRCDIR)/fossil.dom.js \
  $(SRCDIR)/fossil.fetch.js \

  $(SRCDIR)/fossil.wikiedit-wysiwyg.js \
  $(SRCDIR)/graph.js \
  $(SRCDIR)/hbmenu.js \
  $(SRCDIR)/href.js \
  $(SRCDIR)/login.js \
  $(SRCDIR)/markdown.md \
  $(SRCDIR)/menu.js \

  $(SRCDIR)/scroll.js \
  $(SRCDIR)/skin.js \
  $(SRCDIR)/sorttable.js \
  $(SRCDIR)/sounds/0.wav \
  $(SRCDIR)/sounds/1.wav \
  $(SRCDIR)/sounds/2.wav \
  $(SRCDIR)/sounds/3.wav \

    CX("<h2>Wiki Name Rules</h2>");
    well_formed_wiki_name_rules();
    CX("</div>"/*#wikiedit-tab-save*/);
  }
  builtin_fossil_js_bundle_or("fetch", "dom", "tabs", "confirmer",
                              "storage", "popupwidget", "copybutton",
                              "pikchr", NULL);
  builtin_request_js("diff.js");
  builtin_request_js("fossil.page.wikiedit.js");
  builtin_fulfill_js_requests();
  /* Dynamically populate the editor... */
  style_script_begin(__FILE__,__LINE__);
  {
    /* Render the current page list to save us an XHR request
       during page initialization. This must be OUTSIDE of

  *  <a href="../../../fdiff?v1=d1c60722e0b9d775&v2=58d1a8991bacb113"
     target="testwindow">Column alignment with multibyte characters.</a>
     The edit of a line with multibyte characters is the first chunk.
  *  <a href="../../../fdiff?v1=57b0d8183cab0e3d&v2=37b3ef49d73cdfe6"
     target="testwindow">Large diff of sqlite3.c</a>.  This diff was very
     slow prior to the performance enhancement change [9e15437e97].
  *  <a href="../../../info/bda00cbada#chunk49" target="testwindow">
     A difficult indentation change.</a>  UPDATE:  Notice also the improved
      multi-segment update marks on lines 122648 and 122763 on the new side.
  *  <a href="../../../fdiff?v1=bc8100c9ee01b8c2&v2=1d2acc1a2a65c2bf#chunk42"
      target="testwindow">Inverse of the previous.</a>
  *  <a href="../../../fdiff?v1=955cc67ace8fb622&v2=e2e1c87b86664b45#chunk13"
      target="testwindow">Another tricky indentation.</a>  Notice especially
      lines 59398 and 59407 on the left.
  *  <a href="../../../fdiff?v2=955cc67ace8fb622&v1=e2e1c87b86664b45#chunk13"
      target="testwindow">Inverse of the previous.</a>
  *  <a href="../../../fdiff?v1=955cc67ace8fb622&v2=e2e1c87b86664b45#chunk24"
      target="testwindow">A complex change</a> that is difficult to align, and

  $(SRCDIR)/alerts/bflat3.wav \
  $(SRCDIR)/alerts/bloop.wav \
  $(SRCDIR)/alerts/plunk.wav \
  $(SRCDIR)/chat.js \
  $(SRCDIR)/ci_edit.js \
  $(SRCDIR)/copybtn.js \
  $(SRCDIR)/default.css \
  $(SRCDIR)/diff.js \
  $(SRCDIR)/diff.tcl \
  $(SRCDIR)/forum.js \
  $(SRCDIR)/fossil.bootstrap.js \
  $(SRCDIR)/fossil.confirmer.js \
  $(SRCDIR)/fossil.copybutton.js \
  $(SRCDIR)/fossil.dom.js \
  $(SRCDIR)/fossil.fetch.js \

  $(SRCDIR)/fossil.wikiedit-wysiwyg.js \
  $(SRCDIR)/graph.js \
  $(SRCDIR)/hbmenu.js \
  $(SRCDIR)/href.js \
  $(SRCDIR)/login.js \
  $(SRCDIR)/markdown.md \
  $(SRCDIR)/menu.js \

  $(SRCDIR)/scroll.js \
  $(SRCDIR)/skin.js \
  $(SRCDIR)/sorttable.js \
  $(SRCDIR)/sounds/0.wav \
  $(SRCDIR)/sounds/1.wav \
  $(SRCDIR)/sounds/2.wav \
  $(SRCDIR)/sounds/3.wav \

        "$(SRCDIR)\alerts\bflat3.wav" \
        "$(SRCDIR)\alerts\bloop.wav" \
        "$(SRCDIR)\alerts\plunk.wav" \
        "$(SRCDIR)\chat.js" \
        "$(SRCDIR)\ci_edit.js" \
        "$(SRCDIR)\copybtn.js" \
        "$(SRCDIR)\default.css" \
        "$(SRCDIR)\diff.js" \
        "$(SRCDIR)\diff.tcl" \
        "$(SRCDIR)\forum.js" \
        "$(SRCDIR)\fossil.bootstrap.js" \
        "$(SRCDIR)\fossil.confirmer.js" \
        "$(SRCDIR)\fossil.copybutton.js" \
        "$(SRCDIR)\fossil.dom.js" \
        "$(SRCDIR)\fossil.fetch.js" \

        "$(SRCDIR)\fossil.wikiedit-wysiwyg.js" \
        "$(SRCDIR)\graph.js" \
        "$(SRCDIR)\hbmenu.js" \
        "$(SRCDIR)\href.js" \
        "$(SRCDIR)\login.js" \
        "$(SRCDIR)\markdown.md" \
        "$(SRCDIR)\menu.js" \

        "$(SRCDIR)\scroll.js" \
        "$(SRCDIR)\skin.js" \
        "$(SRCDIR)\sorttable.js" \
        "$(SRCDIR)\sounds\0.wav" \
        "$(SRCDIR)\sounds\1.wav" \
        "$(SRCDIR)\sounds\2.wav" \
        "$(SRCDIR)\sounds\3.wav" \

	echo "$(SRCDIR)\alerts/bflat3.wav" >> $@
	echo "$(SRCDIR)\alerts/bloop.wav" >> $@
	echo "$(SRCDIR)\alerts/plunk.wav" >> $@
	echo "$(SRCDIR)\chat.js" >> $@
	echo "$(SRCDIR)\ci_edit.js" >> $@
	echo "$(SRCDIR)\copybtn.js" >> $@
	echo "$(SRCDIR)\default.css" >> $@
	echo "$(SRCDIR)\diff.js" >> $@
	echo "$(SRCDIR)\diff.tcl" >> $@
	echo "$(SRCDIR)\forum.js" >> $@
	echo "$(SRCDIR)\fossil.bootstrap.js" >> $@
	echo "$(SRCDIR)\fossil.confirmer.js" >> $@
	echo "$(SRCDIR)\fossil.copybutton.js" >> $@
	echo "$(SRCDIR)\fossil.dom.js" >> $@
	echo "$(SRCDIR)\fossil.fetch.js" >> $@

	echo "$(SRCDIR)\fossil.wikiedit-wysiwyg.js" >> $@
	echo "$(SRCDIR)\graph.js" >> $@
	echo "$(SRCDIR)\hbmenu.js" >> $@
	echo "$(SRCDIR)\href.js" >> $@
	echo "$(SRCDIR)\login.js" >> $@
	echo "$(SRCDIR)\markdown.md" >> $@
	echo "$(SRCDIR)\menu.js" >> $@

	echo "$(SRCDIR)\scroll.js" >> $@
	echo "$(SRCDIR)\skin.js" >> $@
	echo "$(SRCDIR)\sorttable.js" >> $@
	echo "$(SRCDIR)\sounds/0.wav" >> $@
	echo "$(SRCDIR)\sounds/1.wav" >> $@
	echo "$(SRCDIR)\sounds/2.wav" >> $@
	echo "$(SRCDIR)\sounds/3.wav" >> $@

Notes On Diff Formatting
========================

There are two main kinds of diff display for the web interface:
unified and side-by-side.  Both displays are implemented using
a <table>.  The unified diff is a 4-column table, and the
side-by-side diff is a 5-column table.  In a page like /info that
might show multiple file diffs, each file diff is in a separate
<table>.  For side-by-side diffs, a small amount of Javascript
code is used to resize the text columns so that they fill the screen
width and to keep horizontal scrollbars in sync.

For the unified diff, the basic structure
is like this:

> ~~~~
<table class='diff udiff'>
<tr>
  <td class='diffln difflnl'><pre>
     Line numbers for the left-hand file
  </pre></td>
  <td class='diffln difflnr'><pre>
     Line numbers for the right-hand file
  </pre></td>
  <td class='diffsep'><pre>
     Change marks.  "+" or "=" or nothing
  </pre></td>
  <td class='difftxt difftxtu'><pre>
     The text
  </pre></td>
</tr>
</table>
~~~~

The structure for a side-by-side diff follows the
same basic pattern, though with 5 columns instead of
4, and slightly different class names:

> ~~~~
<table class='diff splitdiff'>
<tr>
  <td class='diffln difflnl'><pre>
     Line numbers for the left-hand file
  </pre></td>
  <td class='difftxt difftxtl'><pre>
     The text for the left side
  </pre></td>
  <td class='diffsep'><pre>
     Change marks.  "+" or "=" or nothing
  </pre></td>
  <td class='diffln difflnr'><pre>
     Line numbers for the right-hand file
  </pre></td>
  <td class='difftxt difftxtr'><pre>
     The text on the right-hand side
  </pre></td>
</tr>
</table>
~~~~

The outer &lt;table&gt; always has class "diff".  The "diffu" class
is added for unified diffs and the "splitdiff" class is added for
side-by-side diffs.

All line-number columns have the "diffln" class.  They also always
have one of "difflnl" or "difflnr" depending on whether they hold
line numbers for the left or right files, respectively.

Text is always kept in a separate column so that it can be scraped
and copied by the user.  All text columns have the "difftxt" class.
One additional class "difftxtu", "difftxtl", or "difftxtr" is added
depending on if the text is for a unified diff, the left column of
a side-by-side diff, or the right column of a side-by-side diff.

The content of all columns is a single &lt;pre&gt; that contains the
appropriate diff-text for that column.  Scrolling is done on the
&lt;pre&gt; element.

Within text columns, highlighting is done with &lt;del&gt; and
&lt;ins&gt; markup.  All text on a line that contains an isert or
delete is surrounded by &lt;ins&gt;...&lt;/ins&gt; or
&lt;del&gt;..&lt;/del&gt;.  Within that line, specific characters
of text that specifically inserted deleted have an additional
layer of &lt;ins&gt; or &lt;del&gt; markup.  Thus CSS like the
following is appropriate:

> ~~~~
td.difftxt ins {
  background-color: #dafbe1;  /* Light green for the whole line */
  text-decoration: none;
}
td.difftxt ins > ins {
  background-color: #a0e4b2;  /* Dark green for specific characters that change */
  text-decoration: none;
}
~~~~

In a side-by-side diff, if an interior &lt;ins&gt; or &lt;del&gt; that mark
specific characters that change correspond to a delete/insert on the other
side, they they have the "edit" class tag.  (ex:  &lt;ins&nbsp;class='edit'&gt;
or &lt;del&nbsp;class='edit'&gt;).  Some skins choose to paint these "modified"
regions blue:

> ~~~~
td.difftxt ins > ins.edit {
  background-color: #c0c0ff;  /* Blue for "modified" text region */
  text-decoration: none;
}
~~~~

Line number text also has &lt;ins&gt; and &lt;del&gt; tags for lines which
are pure insert or pure delete.  But the tags do not nest for line numbers.

Cascading Style Sheet Notes
===========================

This is a collection of technical notes that document the design of
the Document Object Model (DOM) and corresponding Cascading Style Sheet (CSS)
attributes used for customing the look-and-feel of Fossil.  These notes
are of interest to people who want to customize the Fossil skin or
enhance the internal display logic.

This is a collection of documents that we hope will grow over time.

  *  [Diff styling](./diff.md)