Fossil: Check-in [16fde3ff]

Many hyperlinks are disabled.
Use anonymous login to enable hyperlinks.

Overview

Comment:	Add output buffering to the (non-legacy) comment printing algorithm, to reduce calls to fossil_print(). The resulting performance improvement can be up to factor 10, with a perceptible difference even for short comments (measured and tested on Windows with MSVC builds, and on Ubuntu with GCC builds). (For comparison: for the legacy comment printing algorithm, the extra UTF-8 checks added by this branch impair performance by 0.12-1.8%, depending on whether the input contains predominantly multi-byte vs. ASCII-only sequences.)
Downloads:	Tarball \| ZIP archive \| SQL archive
Timelines:	family \| ancestors \| descendants \| both \| comment-formatter-utf8
Files:	files \| file ages \| folders
SHA1:	16fde3ff666cf0733102f7a061756c718597a299
User & Date:	florian 2018-11-15 12:43:00

Context

2018-11-15
15:16		Fix a problem with initial indent introduced by the previous check-in, so that all regression tests from test/comment.test now succeed. Also eliminate three more calls to fossil_print(). Regarding performance, the legacy comment printing algorithm is outnumbered by factor 2-3, with these changes. (check-in: b029ed22 user: florian tags: comment-formatter-utf8)
12:43		Add output buffering to the (non-legacy) comment printing algorithm, to reduce calls to fossil_print(). The resulting performance improvement can be up to factor 10, with a perceptible difference even for short comments (measured and tested on Windows with MSVC builds, and on Ubuntu with GCC builds). (For comparison: for the legacy comment printing algorithm, the extra UTF-8 checks added by this branch impair performance by 0.12-1.8%, depending on whether the input contains predominantly multi-byte vs. ASCII-only sequences.) (check-in: 16fde3ff user: florian tags: comment-formatter-utf8)
2018-10-17
14:16		Modify the comment formatter to avoid output of incomplete UTF-8 sequences, and to avoid line breaks inside UTF-8 sequences. See https://fossil-scm.org/forum/forumpost/1247e4a3c4 for detailed information and tests. (check-in: 1bbca2c3 user: florian tags: comment-formatter-utf8)

Changes

Hide Diffs Unified Diffs Ignore Whitespace Patch

Changes to src/comformat.c.

  int trimSpace,         /* [in] Non-zero to trim leading/trailing spaces. */
  int wordBreak,         /* [in] Non-zero to try breaking on word boundaries. */
  int origBreak,         /* [in] Non-zero to break before original comment. */
  int *pLineCnt,         /* [in/out] Pointer to the total line count. */
  const char **pzLine    /* [out] Pointer to the end of the logical line. */
){
  int index = 0, charCnt = 0, lineCnt = 0, maxChars;

  if( !zLine ) return;
  if( lineChars<=0 ) return;
  comment_print_indent(zLine, indent, trimCrLf, trimSpace, &index);
  maxChars = lineChars;
  for(;;){
    int useChars = 1;
    char c = zLine[index];







    if( c==0 ){
      break;
    }else{
      if( origBreak && index>0 ){
        const char *zCurrent = &zLine[index];
        if( comment_check_orig(zOrigText, zCurrent, &charCnt, &lineCnt) ){






          comment_print_indent(zCurrent, origIndent, trimCrLf, trimSpace,
                               &index);
          maxChars = lineChars;
        }
      }
      index++;
    }
................................................................................
      int nextIndex = comment_next_space(zLine, index);
      if( nextIndex<=0 || (nextIndex-index)>maxChars ){
        break;
      }
      charCnt++;
      useChars = COMMENT_TAB_WIDTH;
      if( maxChars<useChars ){
        fossil_print(" ");
        break;
      }
    }else if( wordBreak && fossil_isspace(c) ){
      int nextIndex = comment_next_space(zLine, index);
      if( nextIndex<=0 || (nextIndex-index)>maxChars ){
        break;
      }
................................................................................
    ** inside UTF-8 sequences. Incomplete, ill-formed and overlong sequences are
    ** kept together. The invalid lead bytes 0xC0 to 0xC1 and 0xF5 to 0xF7 are
    ** allowed to initiate (ill-formed) 2- and 4-byte sequences, respectively,
    ** the other invalid lead bytes 0xF8 to 0xFF are treated as invalid 1-byte
    ** sequences (as lone trail bytes).
    */
    if( (c&0xc0)==0xc0 && zLine[index]!=0 ){  /* Any UTF-8 lead byte 11xxxxxx */
      char zUTF8[5]; /* Buffer to hold a UTF-8 sequence. */
      int cchUTF8=1; /* Code units consumed. */
      int maxUTF8=1; /* Expected sequence length. */
      zUTF8[0]=c;

      if( (c&0xe0)==0xc0 )maxUTF8=2;          /* UTF-8 lead byte 110vvvvv */
      else if( (c&0xf0)==0xe0 )maxUTF8=3;     /* UTF-8 lead byte 1110vvvv */
      else if( (c&0xf8)==0xf0 )maxUTF8=4;     /* UTF-8 lead byte 11110vvv */
      while( cchUTF8<maxUTF8 &&
              (zLine[index]&0xc0)==0x80 ){    /* UTF-8 trail byte 10vvvvvv */

        zUTF8[cchUTF8++] = zLine[index++];
      }
      zUTF8[cchUTF8]=0;
      fossil_print("%s", zUTF8);
    }
    else
      fossil_print("%c", c);

    if( (c&0x80)==0 || (zLine[index+1]&0xc0)!=0xc0 ) maxChars -= useChars;
    if( maxChars<=0 ) break;
    if( c=='\n' ) break;
  }
  if( charCnt>0 ){
    fossil_print("\n");
    lineCnt++;






  }
  if( pLineCnt ){
    *pLineCnt += lineCnt;
  }
  if( pzLine ){
    *pzLine = zLine + index;
  }








>







>
>
>
>
>
>
>






>
>
>
>
>
>







 







|







 







<


<
>





>
|

<
<


<
>





|

>
>
>
>
>
>

  int trimSpace,         /* [in] Non-zero to trim leading/trailing spaces. */
  int wordBreak,         /* [in] Non-zero to try breaking on word boundaries. */
  int origBreak,         /* [in] Non-zero to break before original comment. */
  int *pLineCnt,         /* [in/out] Pointer to the total line count. */
  const char **pzLine    /* [out] Pointer to the end of the logical line. */
){
  int index = 0, charCnt = 0, lineCnt = 0, maxChars;
  char zBuf[400]; int iBuf=0; /* Output buffer and counter. */
  if( !zLine ) return;
  if( lineChars<=0 ) return;
  comment_print_indent(zLine, indent, trimCrLf, trimSpace, &index);
  maxChars = lineChars;
  for(;;){
    int useChars = 1;
    char c = zLine[index];
    /* Flush the output buffer if there's no space left for at least one more
    ** (potentially 4-byte) UTF-8 sequence and a terminating NULL. */
    if ( iBuf>sizeof(zBuf)-5 ){
      zBuf[iBuf]=0;
      iBuf=0;
      fossil_print("%s", zBuf);
    }
    if( c==0 ){
      break;
    }else{
      if( origBreak && index>0 ){
        const char *zCurrent = &zLine[index];
        if( comment_check_orig(zOrigText, zCurrent, &charCnt, &lineCnt) ){
          /* Flush the output buffer before printing the indentation. */
          if ( iBuf>0 ){
            zBuf[iBuf]=0;
            iBuf=0;
            fossil_print("%s", zBuf);
          }
          comment_print_indent(zCurrent, origIndent, trimCrLf, trimSpace,
                               &index);
          maxChars = lineChars;
        }
      }
      index++;
    }
................................................................................
      int nextIndex = comment_next_space(zLine, index);
      if( nextIndex<=0 || (nextIndex-index)>maxChars ){
        break;
      }
      charCnt++;
      useChars = COMMENT_TAB_WIDTH;
      if( maxChars<useChars ){
        zBuf[iBuf++] = ' ';
        break;
      }
    }else if( wordBreak && fossil_isspace(c) ){
      int nextIndex = comment_next_space(zLine, index);
      if( nextIndex<=0 || (nextIndex-index)>maxChars ){
        break;
      }
................................................................................
    ** inside UTF-8 sequences. Incomplete, ill-formed and overlong sequences are
    ** kept together. The invalid lead bytes 0xC0 to 0xC1 and 0xF5 to 0xF7 are
    ** allowed to initiate (ill-formed) 2- and 4-byte sequences, respectively,
    ** the other invalid lead bytes 0xF8 to 0xFF are treated as invalid 1-byte
    ** sequences (as lone trail bytes).
    */
    if( (c&0xc0)==0xc0 && zLine[index]!=0 ){  /* Any UTF-8 lead byte 11xxxxxx */

      int cchUTF8=1; /* Code units consumed. */
      int maxUTF8=1; /* Expected sequence length. */

      zBuf[iBuf++]=c;
      if( (c&0xe0)==0xc0 )maxUTF8=2;          /* UTF-8 lead byte 110vvvvv */
      else if( (c&0xf0)==0xe0 )maxUTF8=3;     /* UTF-8 lead byte 1110vvvv */
      else if( (c&0xf8)==0xf0 )maxUTF8=4;     /* UTF-8 lead byte 11110vvv */
      while( cchUTF8<maxUTF8 &&
              (zLine[index]&0xc0)==0x80 ){    /* UTF-8 trail byte 10vvvvvv */
        cchUTF8++;
        zBuf[iBuf++] = zLine[index++];
      }


    }
    else

      zBuf[iBuf++] = c;
    if( (c&0x80)==0 || (zLine[index+1]&0xc0)!=0xc0 ) maxChars -= useChars;
    if( maxChars<=0 ) break;
    if( c=='\n' ) break;
  }
  if( charCnt>0 ){
    zBuf[iBuf++] = '\n';
    lineCnt++;
  }
  /* Flush the remaining output buffer. */
  if ( iBuf>0 ) {
    zBuf[iBuf]=0;
    iBuf=0;
    fossil_print("%s", zBuf);
  }
  if( pLineCnt ){
    *pLineCnt += lineCnt;
  }
  if( pzLine ){
    *pzLine = zLine + index;
  }