Hi, this patch implements getChars as improvement to the getChar interface,
just by making the number of function calls smaller, i get around 10% speed
improvements in PDF files like 28225 and thought the number of changed lines
is quite big it doesn't add much code duplication (it removes some) so it
should not make code more difficult to maintain.
I've run it thorugh my test suite and did get 100% exact results as the one we
get now.
If noone disagrees i'll commit it to trunk on wednesday european night.
Albert
diff --git a/poppler/Catalog.cc b/poppler/Catalog.cc
index 900cdd7..dbf9af2 100644
--- a/poppler/Catalog.cc
+++ b/poppler/Catalog.cc
@@ -192,7 +192,6 @@ GooString *Catalog::readMetadata() {
GooString *s;
Dict *dict;
Object obj;
- int c;
if (metadata.isNone()) {
Object catDict;
@@ -217,10 +216,7 @@ GooString *Catalog::readMetadata() {
}
obj.free();
s = new GooString();
- metadata.streamReset();
- while ((c = metadata.streamGetChar()) != EOF) {
- s->append(c);
- }
+ metadata.getStream()->fillGooString(s);
metadata.streamClose();
return s;
}
@@ -409,11 +405,7 @@ GooString *Catalog::getJS(int i)
else if (obj2.isStream()) {
Stream *stream = obj2.getStream();
js = new GooString();
- stream->reset();
- int j;
- while ((j = stream->getChar()) != EOF) {
- js->append((char)j);
- }
+ stream->fillGooString(js);
}
obj2.free();
obj.free();
diff --git a/poppler/DCTStream.cc b/poppler/DCTStream.cc
index 5cbaced..1b15619 100644
--- a/poppler/DCTStream.cc
+++ b/poppler/DCTStream.cc
@@ -145,27 +145,48 @@ void DCTStream::reset() {
}
}
+// we can not go with inline since gcc
+// refuses to inline because of setjmp
+#define DO_GET_CHAR \
+ if (current == limit) { \
+ if (cinfo.output_scanline < cinfo.output_height) \
+ { \
+ if (!setjmp(src.setjmp_buffer)) \
+ { \
+ if (!jpeg_read_scanlines(&cinfo, row_buffer, 1)) c = EOF; \
+ else { \
+ current = &row_buffer[0][0]; \
+ limit = &row_buffer[0][(cinfo.output_width - 1) * cinfo.output_components] + cinfo.output_components; \
+ c = *current; \
+ ++current; \
+ } \
+ } \
+ else c = EOF; \
+ } \
+ else c = EOF; \
+ } else { \
+ c = *current; \
+ ++current; \
+ } \
+
int DCTStream::getChar() {
int c;
- if (current == limit) {
- if (cinfo.output_scanline < cinfo.output_height)
- {
- if (!setjmp(src.setjmp_buffer))
- {
- if (!jpeg_read_scanlines(&cinfo, row_buffer, 1)) return EOF;
- current = &row_buffer[0][0];
- limit = &row_buffer[0][(cinfo.output_width - 1) * cinfo.output_components] + cinfo.output_components;
- }
- else return EOF;
- }
- else return EOF;
- }
- c = *current;
- ++current;
+ DO_GET_CHAR
+
return c;
}
+int DCTStream::getChars(int nChars, Guchar *buffer) {
+ int c;
+ for (int i = 0; i < nChars; ++i) {
+ DO_GET_CHAR
+ if (likely(c != EOF)) buffer[i] = c;
+ else return i;
+ }
+ return nChars;
+}
+
int DCTStream::lookChar() {
return *current;
}
diff --git a/poppler/DCTStream.h b/poppler/DCTStream.h
index 6768ff2..65196ec 100644
--- a/poppler/DCTStream.h
+++ b/poppler/DCTStream.h
@@ -69,6 +69,9 @@ public:
private:
void init();
+ virtual GBool hasGetChars() { return true; }
+ virtual int getChars(int nChars, Guchar *buffer);
+
JSAMPLE *current;
JSAMPLE *limit;
struct jpeg_decompress_struct cinfo;
diff --git a/poppler/GfxFont.cc b/poppler/GfxFont.cc
index f823faf..4764b84 100644
--- a/poppler/GfxFont.cc
+++ b/poppler/GfxFont.cc
@@ -421,17 +421,13 @@ CharCodeToUnicode *GfxFont::readToUnicodeCMap(Dict *fontDict, int nBits,
CharCodeToUnicode *ctu) {
GooString *buf;
Object obj1;
- int c;
if (!fontDict->lookup("ToUnicode", &obj1)->isStream()) {
obj1.free();
return NULL;
}
buf = new GooString();
- obj1.streamReset();
- while ((c = obj1.streamGetChar()) != EOF) {
- buf->append(c);
- }
+ obj1.getStream()->fillGooString(buf);
obj1.streamClose();
obj1.free();
if (ctu) {
@@ -488,8 +484,6 @@ char *GfxFont::readEmbFontFile(XRef *xref, int *len) {
char *buf;
Object obj1, obj2;
Stream *str;
- int c;
- int size, i;
obj1.initRef(embFontID.num, embFontID.gen);
obj1.fetch(xref, &obj2);
@@ -503,17 +497,7 @@ char *GfxFont::readEmbFontFile(XRef *xref, int *len) {
}
str = obj2.getStream();
- buf = NULL;
- i = size = 0;
- str->reset();
- while ((c = str->getChar()) != EOF) {
- if (i == size) {
- size += 4096;
- buf = (char *)grealloc(buf, size);
- }
- buf[i++] = c;
- }
- *len = i;
+ buf = (char*)str->toUnsignedChars(len);
str->close();
obj2.free();
diff --git a/poppler/GfxState.cc b/poppler/GfxState.cc
index 12e14d6..5e0be99 100644
--- a/poppler/GfxState.cc
+++ b/poppler/GfxState.cc
@@ -1481,22 +1481,11 @@ GfxColorSpace *GfxICCBasedColorSpace::parse(Array *arr, Gfx *gfx) {
arr->get(1, &obj1);
dict = obj1.streamGetDict();
Guchar *profBuf;
- unsigned int bufSize;
Stream *iccStream = obj1.getStream();
- int c;
- unsigned int size = 0;
-
- bufSize = 65536;
- profBuf = (Guchar *)gmallocn(bufSize,1);
- iccStream->reset();
- while ((c = iccStream->getChar()) != EOF) {
- if (bufSize <= size) {
- bufSize += 65536;
- profBuf = (Guchar *)greallocn(profBuf,bufSize,1);
- }
- profBuf[size++] = c;
- }
- cmsHPROFILE hp = cmsOpenProfileFromMem(profBuf,size);
+ int length = 0;
+
+ profBuf = iccStream->toUnsignedChars(&length, 65536, 65536);
+ cmsHPROFILE hp = cmsOpenProfileFromMem(profBuf,length);
gfree(profBuf);
if (hp == 0) {
error(-1, "read ICCBased color space profile error");
@@ -1715,7 +1704,6 @@ GfxColorSpace *GfxIndexedColorSpace::parse(Array *arr, Gfx *gfx) {
GfxColorSpace *baseA;
int indexHighA;
Object obj1;
- int x;
char *s;
int n, i, j;
@@ -1752,12 +1740,10 @@ GfxColorSpace *GfxIndexedColorSpace::parse(Array *arr, Gfx *gfx) {
if (obj1.isStream()) {
obj1.streamReset();
for (i = 0; i <= indexHighA; ++i) {
- for (j = 0; j < n; ++j) {
- if ((x = obj1.streamGetChar()) == EOF) {
- error(-1, "Bad Indexed color space (lookup table stream too short) padding with zeroes");
- x = 0;
- }
- cs->lookup[i*n + j] = (Guchar)x;
+ const int readChars = obj1.streamGetChars(n, &cs->lookup[i*n]);
+ for (j = readChars; j < n; ++j) {
+ error(-1, "Bad Indexed color space (lookup table stream too short) padding with zeroes");
+ cs->lookup[i*n + j] = 0;
}
}
obj1.streamClose();
diff --git a/poppler/JPEG2000Stream.cc b/poppler/JPEG2000Stream.cc
index 3b301f7..c8bf2c9 100644
--- a/poppler/JPEG2000Stream.cc
+++ b/poppler/JPEG2000Stream.cc
@@ -43,42 +43,34 @@ int JPXStream::getPos() {
return counter;
}
+int JPXStream::getChars(int nChars, Guchar *buffer) {
+ for (int i = 0; i < nChars; ++i) {
+ const int c = doGetChar();
+ if (likely(c != EOF)) buffer[i] = c;
+ else return i;
+ }
+ return nChars;
+}
+
int JPXStream::getChar() {
- int result = lookChar();
- ++counter;
- return result;
+ return doGetChar();
}
-#define BUFFER_INCREASE 4096
+#define BUFFER_INITIAL_SIZE 4096
void JPXStream::init()
{
Object oLen;
if (getDict()) getDict()->lookup("Length", &oLen);
- int bufSize = BUFFER_INCREASE;
+ int bufSize = BUFFER_INITIAL_SIZE;
if (oLen.isInt()) bufSize = oLen.getInt();
oLen.free();
- unsigned char *buf = (unsigned char*)gmallocn(bufSize, sizeof(unsigned char));
- int index = 0;
-
- str->reset();
- int c = str->getChar();
- while(c != EOF)
- {
- if (index >= bufSize)
- {
- bufSize += BUFFER_INCREASE;
- buf = (unsigned char*)greallocn(buf, bufSize, sizeof(unsigned char));
- }
- buf[index] = c;
- ++index;
- c = str->getChar();
- }
-
- init2(buf, index, CODEC_JP2);
-
+
+ int length = 0;
+ unsigned char *buf = str->toUnsignedChars(&length, bufSize);
+ init2(buf, length, CODEC_JP2);
free(buf);
counter = 0;
@@ -143,30 +135,7 @@ error:
}
int JPXStream::lookChar() {
- if (inited == gFalse) init();
-
- if (!image) return EOF;
-
- int w = image->comps[0].w;
- int h = image->comps[0].h;
-
- int y = (counter / image->numcomps) / w;
- int x = (counter / image->numcomps) % w;
- if (y >= h) return EOF;
-
- int component = counter % image->numcomps;
-
- int adjust = 0;
- if (image->comps[component].prec > 8) {
- adjust = image->comps[component].prec - 8;
- }
-
- int r = image->comps[component].data[y * w + x];
- r += (image->comps[component].sgnd ? 1 << (image->comps[0].prec - 1) : 0);
-
- unsigned char rc = (unsigned char) ((r >> adjust)+((r >> (adjust-1))%2));
-
- return rc;
+ return doLookChar();
}
GooString *JPXStream::getPSFilter(int psLevel, char *indent) {
diff --git a/poppler/JPEG2000Stream.h b/poppler/JPEG2000Stream.h
index ea32093..df9a0bf 100644
--- a/poppler/JPEG2000Stream.h
+++ b/poppler/JPEG2000Stream.h
@@ -39,6 +39,42 @@ private:
void init();
void init2(unsigned char *buf, int bufLen, OPJ_CODEC_FORMAT format);
+ virtual GBool hasGetChars() { return true; }
+ virtual int getChars(int nChars, Guchar *buffer);
+
+ inline int doGetChar() {
+ int result = doLookChar();
+ ++counter;
+ return result;
+ }
+
+ inline int doLookChar() {
+ if (inited == gFalse) init();
+
+ if (!image) return EOF;
+
+ int w = image->comps[0].w;
+ int h = image->comps[0].h;
+
+ int y = (counter / image->numcomps) / w;
+ int x = (counter / image->numcomps) % w;
+ if (y >= h) return EOF;
+
+ int component = counter % image->numcomps;
+
+ int adjust = 0;
+ if (image->comps[component].prec > 8) {
+ adjust = image->comps[component].prec - 8;
+ }
+
+ int r = image->comps[component].data[y * w + x];
+ r += (image->comps[component].sgnd ? 1 << (image->comps[0].prec - 1) : 0);
+
+ unsigned char rc = (unsigned char) ((r >> adjust)+((r >> (adjust-1))%2));
+
+ return rc;
+ }
+
opj_image_t *image;
opj_dinfo_t *dinfo;
int counter;
diff --git a/poppler/Link.cc b/poppler/Link.cc
index 5d7b779..72450cc 100644
--- a/poppler/Link.cc
+++ b/poppler/Link.cc
@@ -704,11 +704,7 @@ LinkRendition::LinkRendition(Object *obj) {
} else if (tmp.isStream()) {
Stream *stream = tmp.getStream();
js = new GooString();
- stream->reset();
- int i;
- while ((i = stream->getChar()) != EOF) {
- js->append((char)i);
- }
+ stream->fillGooString(js);
} else {
error(-1, "Invalid Rendition Action: JS not string or stream");
}
@@ -766,11 +762,7 @@ LinkJavaScript::LinkJavaScript(Object *jsObj) {
else if (jsObj->isStream()) {
Stream *stream = jsObj->getStream();
js = new GooString();
- stream->reset();
- int i;
- while ((i = stream->getChar()) != EOF) {
- js->append((char)i);
- }
+ stream->fillGooString(js);
}
}
diff --git a/poppler/Object.h b/poppler/Object.h
index 2b9f20c..60bac4f 100644
--- a/poppler/Object.h
+++ b/poppler/Object.h
@@ -230,6 +230,7 @@ public:
void streamReset();
void streamClose();
int streamGetChar();
+ int streamGetChars(int nChars, Guchar *buffer);
int streamLookChar();
char *streamGetLine(char *buf, int size);
Guint streamGetPos();
@@ -341,6 +342,9 @@ inline void Object::streamClose()
inline int Object::streamGetChar()
{ OBJECT_TYPE_CHECK(objStream); return stream->getChar(); }
+inline int Object::streamGetChars(int nChars, Guchar *buffer)
+ { OBJECT_TYPE_CHECK(objStream); return stream->doGetChars(nChars, buffer); }
+
inline int Object::streamLookChar()
{ OBJECT_TYPE_CHECK(objStream); return stream->lookChar(); }
diff --git a/poppler/Stream.cc b/poppler/Stream.cc
index 0771e25..d6efc74 100644
--- a/poppler/Stream.cc
+++ b/poppler/Stream.cc
@@ -99,6 +99,15 @@ int Stream::getRawChar() {
return EOF;
}
+int Stream::getChars(int nChars, Guchar *buffer) {
+ error(-1, "Internal: called getChars() on non-predictor stream");
+ return 0;
+}
+
+void Stream::getRawChars(int nChars, int *buffer) {
+ error(-1, "Internal: called getRawChars() on non-predictor stream");
+}
+
char *Stream::getLine(char *buf, int size) {
int i;
int c;
@@ -461,9 +470,8 @@ Guchar *ImageStream::getLine() {
}
} else if (nBits == 8) {
Guchar *line = imgLine;
- for (i = 0; i < nVals; ++i) {
- *line++ = str->getChar();
- }
+ int readChars = str->doGetChars(nVals, line);
+ for ( ; readChars < nVals; readChars++) line[readChars] = EOF;
} else if (nBits == 16) {
// this is a hack to support 16 bits images, everywhere
// we assume a component fits in 8 bits, with this hack
@@ -543,12 +551,17 @@ int StreamPredictor::lookChar() {
}
int StreamPredictor::getChar() {
- if (predIdx >= rowBytes) {
- if (!getNextLine()) {
- return EOF;
- }
+ return doGetChar();
+}
+
+int StreamPredictor::getChars(int nChars, Guchar *buffer)
+{
+ for (int i = 0; i < nChars; ++i) {
+ const int c = doGetChar();
+ if (likely(c != EOF)) buffer[i] = c;
+ else return i;
}
- return predLine[predIdx++];
+ return nChars;
}
GBool StreamPredictor::getNextLine() {
@@ -571,13 +584,15 @@ GBool StreamPredictor::getNextLine() {
}
// read the raw line, apply PNG (byte) predictor
+ int *rawCharLine = new int[rowBytes - pixBytes];
+ str->getRawChars(rowBytes - pixBytes, rawCharLine);
memset(upLeftBuf, 0, pixBytes + 1);
for (i = pixBytes; i < rowBytes; ++i) {
for (j = pixBytes; j > 0; --j) {
upLeftBuf[j] = upLeftBuf[j-1];
}
upLeftBuf[0] = predLine[i];
- if ((c = str->getRawChar()) == EOF) {
+ if ((c = rawCharLine[i - pixBytes]) == EOF) {
if (i > pixBytes) {
// this ought to return false, but some (broken) PDF files
// contain truncated image data, and Adobe apparently reads the
@@ -621,6 +636,7 @@ GBool StreamPredictor::getNextLine() {
break;
}
}
+ delete[] rawCharLine;
// apply TIFF (component) predictor
if (predictor == 2) {
@@ -1237,16 +1253,13 @@ int LZWStream::lookChar() {
return seqBuf[seqIndex];
}
+void LZWStream::getRawChars(int nChars, int *buffer) {
+ for (int i = 0; i < nChars; ++i)
+ buffer[i] = doGetRawChar();
+}
+
int LZWStream::getRawChar() {
- if (eof) {
- return EOF;
- }
- if (seqIndex >= seqLength) {
- if (!processNextCode()) {
- return EOF;
- }
- }
- return seqBuf[seqIndex++];
+ return doGetRawChar();
}
void LZWStream::reset() {
@@ -4231,20 +4244,20 @@ void FlateStream::reset() {
}
int FlateStream::getChar() {
- int c;
+ return doGetChar();
+}
+int FlateStream::getChars(int nChars, Guchar *buffer) {
if (pred) {
- return pred->getChar();
- }
- while (remain == 0) {
- if (endOfBlock && eof)
- return EOF;
- readSome();
+ return pred->getChars(nChars, buffer);
+ } else {
+ for (int i = 0; i < nChars; ++i) {
+ const int c = doGetChar();
+ if (likely(c != EOF)) buffer[i] = c;
+ else return i;
+ }
+ return nChars;
}
- c = buf[index];
- index = (index + 1) & flateMask;
- --remain;
- return c;
}
int FlateStream::lookChar() {
@@ -4262,18 +4275,13 @@ int FlateStream::lookChar() {
return c;
}
-int FlateStream::getRawChar() {
- int c;
+void FlateStream::getRawChars(int nChars, int *buffer) {
+ for (int i = 0; i < nChars; ++i)
+ buffer[i] = doGetRawChar();
+}
- while (remain == 0) {
- if (endOfBlock && eof)
- return EOF;
- readSome();
- }
- c = buf[index];
- index = (index + 1) & flateMask;
- --remain;
- return c;
+int FlateStream::getRawChar() {
+ return doGetRawChar();
}
GooString *FlateStream::getPSFilter(int psLevel, char *indent) {
diff --git a/poppler/Stream.h b/poppler/Stream.h
index 49ae8fb..4f7be01 100644
--- a/poppler/Stream.h
+++ b/poppler/Stream.h
@@ -106,6 +106,56 @@ public:
// Close down the stream.
virtual void close();
+ inline int doGetChars(int nChars, Guchar *buffer)
+ {
+ if (hasGetChars()) {
+ return getChars(nChars, buffer);
+ } else {
+ for (int i = 0; i < nChars; ++i) {
+ const int c = getChar();
+ if (likely(c != EOF)) buffer[i] = c;
+ else return i;
+ }
+ return nChars;
+ }
+ }
+
+ inline void fillGooString(GooString *s)
+ {
+ Guchar readBuf[4096];
+ int readChars;
+ reset();
+ while ((readChars = doGetChars(4096, readBuf)) != 0) {
+ s->append((const char *)readBuf, readChars);
+ }
+ }
+
+ inline Guchar *toUnsignedChars(int *length, int initialSize = 4096, int sizeIncrement = 4096)
+ {
+ int readChars;
+ Guchar *buf = (Guchar *)gmalloc(initialSize);
+ int size = initialSize;
+ *length = 0;
+ int charsToRead = initialSize;
+ bool continueReading = true;
+ reset();
+ while (continueReading && (readChars = doGetChars(charsToRead, &buf[*length])) != 0) {
+ *length += readChars;
+ if (readChars == charsToRead) {
+ if (lookChar() != EOF) {
+ size += sizeIncrement;
+ charsToRead = initialSize;
+ buf = (Guchar *)grealloc(buf, size);
+ } else {
+ continueReading = false;
+ }
+ } else {
+ continueReading = false;
+ }
+ }
+ return buf;
+ }
+
// Get next char from stream.
virtual int getChar() = 0;
@@ -115,6 +165,7 @@ public:
// Get next char from stream without using the predictor.
// This is only used by StreamPredictor.
virtual int getRawChar();
+ virtual void getRawChars(int nChars, int *buffer);
// Get next char directly from stream source, without filtering it
virtual int getUnfilteredChar () = 0;
@@ -166,6 +217,8 @@ public:
Stream *addFilters(Object *dict);
private:
+ virtual GBool hasGetChars() { return false; }
+ virtual int getChars(int nChars, Guchar *buffer);
Stream *makeFilter(char *name, Stream *str, Object *params);
@@ -347,11 +400,21 @@ public:
int lookChar();
int getChar();
+ int getChars(int nChars, Guchar *buffer);
private:
GBool getNextLine();
+ inline int doGetChar() {
+ if (predIdx >= rowBytes) {
+ if (!getNextLine()) {
+ return EOF;
+ }
+ }
+ return predLine[predIdx++];
+ }
+
Stream *str; // base stream
int predictor; // predictor
int width; // pixels per line
@@ -383,7 +446,7 @@ public:
virtual void reset();
virtual void close();
virtual int getChar()
- { return (bufPtr >= bufEnd && !fillBuf()) ? EOF : (*bufPtr++ & 0xff); }
+ { return doGetChar(); }
virtual int lookChar()
{ return (bufPtr >= bufEnd && !fillBuf()) ? EOF : (*bufPtr & 0xff); }
virtual int getPos() { return bufPos + (bufPtr - buf); }
@@ -397,6 +460,20 @@ public:
private:
GBool fillBuf();
+
+ inline int doGetChar()
+ { return (bufPtr >= bufEnd && !fillBuf()) ? EOF : (*bufPtr++ & 0xff); }
+
+ virtual GBool hasGetChars() { return true; }
+ virtual int getChars(int nChars, Guchar *buffer)
+ {
+ for (int i = 0; i < nChars; ++i) {
+ const int c = doGetChar();
+ if (likely(c != EOF)) buffer[i] = c;
+ else return i;
+ }
+ return nChars;
+ }
FILE *f;
Guint start;
@@ -596,11 +673,24 @@ public:
virtual int getChar();
virtual int lookChar();
virtual int getRawChar();
+ virtual void getRawChars(int nChars, int *buffer);
virtual GooString *getPSFilter(int psLevel, char *indent);
virtual GBool isBinary(GBool last = gTrue);
private:
+ inline int doGetRawChar() {
+ if (eof) {
+ return EOF;
+ }
+ if (seqIndex >= seqLength) {
+ if (!processNextCode()) {
+ return EOF;
+ }
+ }
+ return seqBuf[seqIndex++];
+ }
+
StreamPredictor *pred; // predictor
int early; // early parameter
GBool eof; // true if at eof
@@ -855,11 +945,35 @@ public:
virtual int getChar();
virtual int lookChar();
virtual int getRawChar();
+ virtual void getRawChars(int nChars, int *buffer);
virtual GooString *getPSFilter(int psLevel, char *indent);
virtual GBool isBinary(GBool last = gTrue);
virtual void unfilteredReset ();
private:
+ inline int doGetRawChar() {
+ int c;
+
+ while (remain == 0) {
+ if (endOfBlock && eof)
+ return EOF;
+ readSome();
+ }
+ c = buf[index];
+ index = (index + 1) & flateMask;
+ --remain;
+ return c;
+ }
+
+ inline int doGetChar() {
+ if (pred) {
+ return pred->getChar();
+ }
+ return doGetRawChar();
+ }
+
+ virtual GBool hasGetChars() { return true; }
+ virtual int getChars(int nChars, Guchar *buffer);
StreamPredictor *pred; // predictor
Guchar buf[flateWindow]; // output data buffer
_______________________________________________
poppler mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/poppler
