Fossil

/*
** A single file change record.
*/
struct ImportFile {
  char *zName;           /* Name of a file */
  char *zUuid;           /* UUID of the file */
  char *zPrior;          /* Prior name if the name was changed */

  char isExe;            /* True if executable */
  char isLink;           /* True if symlink */
  char hasChanged;       /* True when different than baseline */
};
#endif


/*
** State information about an on-going fast-import parse.
*/
static struct {
  void (*xFinish)(void);      /* Function to finish a prior record */
  int nData;                  /* Bytes of data */
  char *zBaseline;            /* Baseline manifest.  The B card. */
  char *zTag;                 /* Name of a tag */
  char *zBranch;              /* Name of a branch for a commit */
  char *zPrevBranch;          /* The branch of the previous check-in */
  char *aData;                /* Data content */
  char *zMark;                /* The current mark */
  char *zDate;                /* Date/time stamp */
  char *zUser;                /* User name */
  char *zComment;             /* Comment of a commit */
  char *zFrom;                /* from value as a UUID */
  char *zPrevCheckin;         /* Name of the previous check-in */
  char *zFromMark;            /* The mark of the "from" field */
  int nMerge;                 /* Number of merge values */
  int nMergeAlloc;            /* Number of slots in azMerge[] */
  char **azMerge;             /* Merge values */
  int nFile;                  /* Number of aFile values */
  int nFileAlloc;             /* Number of slots in aFile[] */
  int nFileEffective;         /* Number of aFile items with zUuid != 0 */
  int nChanged;               /* Number of aFile that differ from baseline */
  ImportFile *aFile;          /* Information about files in a commit */
  int fromLoaded;             /* True zFrom content loaded into aFile[] */
  int hasLinks;               /* True if git repository contains symlinks */
  int tagCommit;              /* True if the commit adds a tag */
  int tryDelta;               /* Attempt to generate delta manifests */
} gg;

/*
** Duplicate a string.
*/
char *fossil_strdup(const char *zOrig){
  char *z = 0;

**
** The azMerge[] and aFile[] arrays are zeroed by allocated space is
** retained unless the freeAll flag is set.
*/
static void import_reset(int freeAll){
  int i;
  gg.xFinish = 0;
  fossil_free(gg.zBaseline); gg.zBaseline = 0;
  fossil_free(gg.zTag); gg.zTag = 0;
  fossil_free(gg.zBranch); gg.zBranch = 0;
  fossil_free(gg.aData); gg.aData = 0;
  fossil_free(gg.zMark); gg.zMark = 0;
  fossil_free(gg.zDate); gg.zDate = 0;
  fossil_free(gg.zUser); gg.zUser = 0;
  fossil_free(gg.zComment); gg.zComment = 0;
  fossil_free(gg.zFrom); gg.zFrom = 0;
  fossil_free(gg.zFromMark); gg.zFromMark = 0;
  for(i=0; i<gg.nMerge; i++){
    fossil_free(gg.azMerge[i]); gg.azMerge[i] = 0;
  }
  gg.nMerge = 0;
  for(i=0; i<gg.nFile; i++){
    fossil_free(gg.aFile[i].zName);
    fossil_free(gg.aFile[i].zUuid);
    fossil_free(gg.aFile[i].zPrior);
    gg.aFile[i].zPrior = 0;
  }
  memset(gg.aFile, 0, gg.nFile*sizeof(gg.aFile[0]));
  gg.nChanged = 0;
  gg.nFileEffective = 0;
  gg.nFile = 0;
  if( freeAll ){
    fossil_free(gg.zPrevBranch);
    fossil_free(gg.zPrevCheckin);
    fossil_free(gg.azMerge);
    fossil_free(gg.aFile);
    memset(&gg, 0, sizeof(gg));

/*
** Use data accumulated in gg from a "commit" record to add a new
** manifest artifact to the BLOB table.
*/
static void finish_commit(void){
  int i;
  int delta;
  char *zFromBranch;
  char *aTCard[4];                /* Array of T cards for manifest */
  int nTCard = 0;                 /* Entries used in aTCard[] */
  Blob record, cksum;

  import_prior_files();
  qsort(gg.aFile, gg.nFile, sizeof(gg.aFile[0]), mfile_cmp);
  /*
  ** This is the same mechanism used by the commit command to decide whether to
  ** generate a delta manifest or not.  It is evaluated and saved for later
  ** uses.
  */
  delta = gg.tryDelta && (gg.zBaseline!=0 || gg.zFrom!=0) &&
                         (gg.nChanged*gg.nChanged)<(gg.nFileEffective*3-9);
  blob_zero(&record);
  if( delta ){
    blob_appendf(&record, "B %s\n", gg.zBaseline!=0 ? gg.zBaseline : gg.zFrom);
  }
  blob_appendf(&record, "C %F\n", gg.zComment);
  blob_appendf(&record, "D %s\n", gg.zDate);
  for(i=0; i<gg.nFile; i++){
    const struct ImportFile *pFile = &gg.aFile[i];
    if( (!delta && pFile->zUuid==0) || (delta && !pFile->hasChanged) ) continue;
    blob_appendf(&record, "F %F", pFile->zName);
    if( pFile->zUuid!=0 ) {
      blob_appendf(&record, " %s", pFile->zUuid);
      if( pFile->isExe ){
        blob_append(&record, " x", 2);
      }else if( pFile->isLink ){
        blob_append(&record, " l", 2);
      }
      if( pFile->zPrior!=0 ){
        if( !pFile->isExe && !pFile->isLink ){

          blob_append(&record, " w", 2);
        }
        blob_appendf(&record, " %F", pFile->zPrior);
      }
    }
    blob_append(&record, "\n", 1);
  }
  if( gg.zFrom ){
    blob_appendf(&record, "P %s", gg.zFrom);
    for(i=0; i<gg.nMerge; i++){
      blob_appendf(&record, " %s", gg.azMerge[i]);
    }
    blob_append(&record, "\n", 1);

** Get a token from a line of text.  Return a pointer to the first
** character of the token and zero-terminate the token.  Make
** *pzIn point to the first character past the end of the zero
** terminator, or at the zero-terminator at EOL.
*/
static char *next_token(char **pzIn){
  char *z = *pzIn;
  int i, j;
  if( z[0]==0 ) return z;

  if( z[0]=='"' ){
    /* Quoted path name */
    z++;
    for(i=0, j=0; z[i] && z[i]!='"' && z[i]!='\n'; i++, j++){
      if( z[i]=='\\' && z[i+1] ){
        char v, c = z[++i];
        switch( c ){
          case 0:
          case '"':  c = '"';  break;
          case '\\': c = '\\'; break;
          case 'a':  c = '\a'; break;
          case 'b':  c = '\b'; break;
          case 'f':  c = '\f'; break;
          case 'n':  c = '\n'; break;
          case 'r':  c = '\r'; break;
          case 't':  c = '\t'; break;
          case 'v':  c = '\v'; break;
          case '0': case '1': case '2': case '3':
            v = (c - '0') << 6;

            c = z[++i];

            if( c < '0' || c > '7' )
              fossil_fatal("Invalid octal digit '%c' in sequence", c);
            v |= (c - '0') << 3;
            c = z[++i];
            if( c < '0' || c > '7' )
              fossil_fatal("Invalid octal digit '%c' in sequence", c);
            v |= (c - '0');
            c = v;
            break;
          default:
            fossil_fatal("Unrecognized escape sequence \"\\%c\"", c);
        }





        z[j] = c;



      }
    }
    if( z[i]=='"' ) z[i++] = 0;



  }else{
    /* Unquoted path name or generic token */
    for(i=0; z[i] && z[i]!=' ' && z[i]!='\n'; i++){}
  }
  if( z[i] ){
    z[i] = 0;
    *pzIn = &z[i+1];
  }else{
    *pzIn = &z[i];
  }
  return z;
}

  }
  if( gg.zFrom==0 ) return;
  rid = fast_uuid_to_rid(gg.zFrom);
  if( rid==0 ) return;
  p = manifest_get(rid, CFTYPE_MANIFEST);
  if( p==0 ) return;
  manifest_file_rewind(p);
  if( gg.tryDelta && p->pBaseline ){
    /*
    ** The manifest_file_next() iterator skips deletion "F" cards in delta
    ** manifests.  But, in order to build more delta manifests, this information
    ** is necessary because it propagates.  Therefore, in this case, the
    ** manifest has to be traversed "manually".
    */
    Manifest *pB = p->pBaseline;
    gg.zBaseline = fossil_strdup(p->zBaseline);
    while( p->iFile<p->nFile || pB->iFile<pB->nFile ){
      pNew = import_add_file();
      if( p->iFile>=p->nFile ){
        /* No more "F" cards in delta manifest, finish the baseline */
        pOld = &pB->aFile[pB->iFile++];
      }else if( pB->iFile>=pB->nFile ){
        /* No more "F" cards in baseline, finish the delta manifest */
        pOld = &p->aFile[p->iFile++];
        pNew->hasChanged = 1;
      }else{
        int cmp = fossil_strcmp(pB->aFile[pB->iFile].zName,
                                p->aFile[p->iFile].zName);
        if( cmp < 0 ){
          pOld = &pB->aFile[pB->iFile++];
        }else if( cmp >= 0 ){
          pOld = &p->aFile[p->iFile++];
          pNew->hasChanged = 1;
          if( cmp==0 ) pB->iFile++;
        }
      }
      pNew->zName = fossil_strdup(pOld->zName);
      pNew->isExe = pOld->zPerm && strstr(pOld->zPerm, "x")!=0;
      pNew->isLink = pOld->zPerm && strstr(pOld->zPerm, "l")!=0;
      pNew->zUuid = fossil_strdup(pOld->zUuid);
      gg.nChanged += pNew->hasChanged;
      if( pNew->zUuid!=0 ) gg.nFileEffective++;
    }
  }else{
    while( (pOld = manifest_file_next(p, 0))!=0 ){
      pNew = import_add_file();
      pNew->zName = fossil_strdup(pOld->zName);
      pNew->isExe = pOld->zPerm && strstr(pOld->zPerm, "x")!=0;
      pNew->isLink = pOld->zPerm && strstr(pOld->zPerm, "l")!=0;
      pNew->zUuid = fossil_strdup(pOld->zUuid);
      if( pNew->zUuid!=0 ) gg.nFileEffective++;
    }
  }
  manifest_destroy(p);
}

/*
** Locate a file in the gg.aFile[] array by its name.  Begin the search
** with the *pI-th file.  Update *pI to be one past the file found.

      return &gg.aFile[i];
    }
    i++;
  }
  return 0;
}



















/*
** Read the git-fast-import format from pIn and insert the corresponding
** content into the database.
*/
static void git_fast_import(FILE *pIn){
  ImportFile *pFile, *pNew;
  int i;
  char *z;
  char *zUuid;
  char *zName;
  char *zPerm;
  char *zFrom;
  char *zTo;
  char zLine[1000];

      if( gg.azMerge[gg.nMerge] ) gg.nMerge++;
    }else
    if( memcmp(zLine, "M ", 2)==0 ){
      import_prior_files();
      z = &zLine[2];
      zPerm = next_token(&z);
      zUuid = next_token(&z);
      zName = next_token(&z);

      i = 0;
      pFile = import_find_file(zName, &i, gg.nFile);
      if( pFile==0 ){
        pFile = import_add_file();
        pFile->zName = fossil_strdup(zName);
        gg.nFileEffective++;
      }
      pFile->isExe = (fossil_strcmp(zPerm, "100755")==0);
      pFile->isLink = (fossil_strcmp(zPerm, "120000")==0);
      fossil_free(pFile->zUuid);
      pFile->zUuid = resolve_committish(zUuid);
      /*
      ** This trick avoids counting multiple changes on the same filename as
      ** changes to multiple filenames.  When an entry in gg.aFile is already
      ** different from its baseline, it is not counted again.
      */
      gg.nChanged += 1 - pFile->hasChanged;
      pFile->hasChanged = 1;
    }else
    if( memcmp(zLine, "D ", 2)==0 ){
      import_prior_files();
      z = &zLine[2];
      zName = next_token(&z);

      i = 0;
      pFile = import_find_file(zName, &i, gg.nFile);
      if( pFile!=0 ){
        /* Do not remove the item from gg.aFile, just mark as deleted */
        fossil_free(pFile->zUuid);
        pFile->zUuid = 0;
        gg.nChanged += 1 - pFile->hasChanged;
        pFile->hasChanged = 1;
        gg.nFileEffective--;
      }
    }else
    if( memcmp(zLine, "C ", 2)==0 ){

      import_prior_files();
      z = &zLine[2];
      zFrom = next_token(&z);
      zTo = next_token(&z);
      i = 0;


      pFile = import_find_file(zFrom, &i, gg.nFile);
      if( pFile!=0 ){
        int j = 0;
        pNew = import_find_file(zTo, &j, gg.nFile);
        if( pNew==0 ){
          pNew = import_add_file();
          pFile = &gg.aFile[i-1];  /* gg.aFile may have been realloc()-ed */

          pNew->zName = fossil_strdup(zTo);
          gg.nFileEffective++;
        }else{
          fossil_free(pNew->zUuid);
        }
        pNew->isExe = pFile->isExe;
        pNew->isLink = pFile->isLink;
        pNew->zUuid = fossil_strdup(pFile->zUuid);
        gg.nChanged += 1 - pNew->hasChanged;
        pNew->hasChanged = 1;
      }
    }else
    if( memcmp(zLine, "R ", 2)==0 ){

      import_prior_files();
      z = &zLine[2];
      zFrom = next_token(&z);
      zTo = next_token(&z);
      i = 0;

      pFile = import_find_file(zFrom, &i, gg.nFile);
      if( pFile!=0 ){
        /*
        ** File renames in delta manifests require two "F" cards: one to
        ** delete the old file (without UUID) and another with the rename
        ** (with prior name equals to the name in the other card).
        **
        ** This forces us to also lookup by the destination name, as it may
        ** already exist in the form of a delta manifest deletion "F" card.
        */
        int j = 0;
        pNew = import_find_file(zTo, &j, gg.nFile);
        if( pNew==0 ){
          pNew = import_add_file();
          pFile = &gg.aFile[i-1];

          pNew->zName = fossil_strdup(zTo);
        }else{
          fossil_free(pNew->zUuid);  /* Just in case */
        }

        pNew->isExe = pFile->isExe;
        pNew->isLink = pFile->isLink;
        pNew->zPrior = fossil_strdup(zFrom);
        pNew->zUuid = pFile->zUuid;
        pFile->zUuid = 0;
        gg.nChanged += 2 - (pNew->hasChanged + pFile->hasChanged);
        pNew->hasChanged = 1;
        pFile->hasChanged = 1;

      }

    }else
    if( memcmp(zLine, "deleteall", 9)==0 ){
      gg.fromLoaded = 1;
    }else
    if( memcmp(zLine, "N ", 2)==0 ){
      /* No-op */
    }else

  fossil_fatal("bad fast-import line: [%s]", zLine);
  return;
}

/*
** COMMAND: import
**
** Usage: %fossil import --git ?OPTIONS? NEW-REPOSITORY ?FILE?
**
** Read text generated by the git-fast-export command and use it to
** construct a new Fossil repository named by the NEW-REPOSITORY
** argument.  If given, the git-fast-export text is read from the FILE argument,
** otherwise text is read from standard input.
**
** The git-fast-export file format is currently the only VCS interchange
** format that is understood, though other interchange formats may be added
** in the future.
**
** The --incremental option allows an existing repository to be extended
** with new content.  Otherwise, if a file with the same name as NEW-REPOSITORY
** is found, the command fails unless the --force option is used.
**
** When the --delta option is used, delta manifests will be generated when they
** are smaller than the equivalent baseline manifest.  Please beware that delta
** manifests are not understood by older versions of Fossil.  Therefore, only
** use this option when it can be assured that only newer clients will pull or
** read from it.
**
** Options:
**   -d|--delta        enable delta manifest generation
**   -f|--force        remove existing file
**   -i|--incremental  allow importing into an existing repository
**
** See also: export
*/
void git_import_cmd(void){
  char *zPassword;
  FILE *pIn;
  Stmt q;
  int forceFlag = find_option("force", "f", 0)!=0;
  int incrFlag = find_option("incremental", "i", 0)!=0;

  find_option("git",0,0);  /* Skip the --git option for now */
  gg.tryDelta = find_option("delta", "d", 0)!=0;
  verify_all_options();
  if( g.argc!=3  && g.argc!=4 ){
    usage("REPOSITORY-NAME");
  }
  if( g.argc==4 ){
    pIn = fossil_fopen(g.argv[3], "rb");
  }else{