--- parser3/src/classes/image.C 2019/12/03 15:09:49 1.166 +++ parser3/src/classes/image.C 2024/12/06 23:20:04 1.194 @@ -1,8 +1,8 @@ /** @file Parser: @b image parser class. - Copyright (c) 2001-2017 Art. Lebedev Studio (http://www.artlebedev.com) - Author: Alexandr Petrosian (http://paf.design.ru) + Copyright (c) 2001-2024 Art. Lebedev Studio (http://www.artlebedev.com) + Authors: Konstantin Morshnev , Alexandr Petrosian */ /* @@ -26,7 +26,7 @@ #include "pa_table.h" #include "pa_charsets.h" -volatile const char * IDENT_IMAGE_C="$Id: image.C,v 1.166 2019/12/03 15:09:49 moko Exp $"; +volatile const char * IDENT_IMAGE_C="$Id: image.C,v 1.194 2024/12/06 23:20:04 moko Exp $"; // defines @@ -189,7 +189,14 @@ public: EXIF_TAG(0xa434, LensModel); EXIF_TAG(0xa435, LensSerialNumber); } -} exif_tag_value2name; + + static EXIF_tag_value2name &instance(){ + static EXIF_tag_value2name *singleton=NULL; + if(!singleton) + singleton=new EXIF_tag_value2name; + return *singleton; + } +}; class EXIF_gps_tag_value2name: public Hash { public: @@ -226,16 +233,25 @@ public: EXIF_TAG(0x1D, GPSDateStamp); EXIF_TAG(0x1E, GPSDifferential); } -} exif_gps_tag_value2name; -#undef EXIF_TAG + static EXIF_gps_tag_value2name &instance(){ + static EXIF_gps_tag_value2name *singleton=NULL; + if(!singleton) + singleton=new EXIF_gps_tag_value2name; + return *singleton; + } + +}; + + +///*********************************************** support functions -#ifndef DOXYGEN class Measure_reader { public: virtual size_t read(const char* &buf, size_t limit)=0; - virtual void seek(long value, int whence)=0; - virtual long tell()=0; + virtual void seek(uint64_t value)=0; + virtual uint64_t tell()=0; + virtual uint64_t length()=0; }; class Measure_file_reader: public Measure_reader { @@ -258,12 +274,14 @@ public: return read_size; } - override void seek(long value, int whence) { - if(lseek(f, value, whence)<0) - throw Exception(IMAGE_FORMAT, &file_name, "seek(value=%ld, whence=%d) failed: %s (%d)", value, whence, strerror(errno), errno); + override void seek(uint64_t value) { + if(pa_lseek(f, value, SEEK_SET)<0) + throw Exception(IMAGE_FORMAT, &file_name, "seek to %.15g failed: %s (%d)", (double)value, strerror(errno), errno); } - override long tell() { return lseek(f, 0, SEEK_CUR); } + override uint64_t tell() { return pa_lseek(f, 0, SEEK_CUR); } + + override uint64_t length() { return pa_lseek(f, 0, SEEK_END); } }; @@ -286,64 +304,48 @@ public: return to_read; } - override void seek(long value, int whence) { - size_t new_offset; - switch(whence) { - case SEEK_CUR: new_offset=offset+value; break; - case SEEK_SET: new_offset=(size_t)value; break; - case SEEK_END: new_offset=size; break; - default: - throw Exception(0, 0, "whence #%d not supported", 0, whence); - break; // never - } - - if((ssize_t)new_offset<0 || new_offset>size) - throw Exception(IMAGE_FORMAT, &file_name, "seek(value=%l, whence=%d) failed: out of buffer, new_offset>size (%l>%l) or new_offset<0", - value, whence, new_offset, size); - offset=new_offset; + override void seek(uint64_t value) { + if(value>(uint64_t)size) + throw Exception(IMAGE_FORMAT, &file_name, "seek to %.15g failed: out of buffer (%.15g)", value, size); + offset=(size_t)value; } - override long tell() { return offset; } + override uint64_t tell() { return offset; } + + override uint64_t length() { return size; } }; -#endif -/// PNG file header -struct PNG_Header { - char dummy[12]; - char signature[4]; //< must be "IHDR" - uchar high_width[2]; //< image width high bytes [we ignore for now] - uchar width[2]; //< image width low bytes - uchar high_height[2]; //< image height high bytes [we ignore for now] - uchar height[4]; //< image height +struct Measure_info { + ushort width; + ushort height; + Value** exif; + Value** xmp; + Charset* xmp_charset; + bool video; }; -/// GIF file header -struct GIF_Header { - char signature[3]; // 'GIF' - char version[3]; - uchar width[2]; - uchar height[2]; - char dif; - char fonColor; - char nulls; -}; -/// BMP file header + DIB header part -struct BMP_Header { - char signature[2]; // 'BM' - uchar file_size[4]; - uchar reserved[4]; - uchar bitmap_offset[4]; - uchar header_size[4]; - uchar width[2]; - uchar high_width[2]; //< image width high bytes [we ignore for now] - uchar height[2]; - uchar high_height[2]; //< image height high bytes [we ignore for now] -}; +inline ushort x_endian_to_ushort(uchar b0, uchar b1) { + return (ushort)((b1<<8) + b0); +} + +inline uint x_endian_to_uint(uchar b0, uchar b1, uchar b2, uchar b3) { + return (uint)(((((b3<<8) + b2)<<8)+b1)<<8)+b0; +} + +inline ushort endian_to_ushort(bool is_big, const uchar *b/* [2] */) { + return is_big ? x_endian_to_ushort(b[1], b[0]) : x_endian_to_ushort(b[0], b[1]); +} + +inline uint endian_to_uint(bool is_big, const uchar *b /* [4] */) { + return is_big ? x_endian_to_uint(b[3], b[2], b[1], b[0]) : x_endian_to_uint(b[0], b[1], b[2], b[3]); +} + + +///*********************************************** JPEG -/// JPEG record head struct JPG_Segment_head { uchar marker; uchar code; @@ -383,37 +385,6 @@ struct JPG_Exif_IFD_entry { #define JPEG_EXIF_DATE_CHARS 20 - -#ifndef DOXYGEN -struct Measure_info { - ushort width; - ushort height; - Value** exif; - Value** xmp; - Charset* xmp_charset; -}; -#endif - -// - -inline ushort x_endian_to_ushort(uchar b0, uchar b1) { - return (ushort)((b1<<8) + b0); -} - -inline uint x_endian_to_uint(uchar b0, uchar b1, uchar b2, uchar b3) { - return (uint)(((((b3<<8) + b2)<<8)+b1)<<8)+b0; -} - -inline ushort endian_to_ushort(bool is_big, const uchar *b/* [2] */) { - return is_big?x_endian_to_ushort(b[1], b[0]): - x_endian_to_ushort(b[0], b[1]); -} - -inline uint endian_to_uint(bool is_big, const uchar *b /* [4] */) { - return is_big?x_endian_to_uint(b[3], b[2], b[1], b[0]): - x_endian_to_uint(b[0], b[1], b[2], b[3]); -} - static Value* parse_IFD_entry_formatted_one_value(bool is_big, ushort format, size_t component_size, const uchar *value) { switch(format) { case 1: // unsigned byte @@ -465,11 +436,8 @@ static Value* parse_IFD_entry_formatted_ size_t length=components_count; // Data format is "YYYY:MM:DD HH:MM:SS"+0x00, total 20bytes if(length==JPEG_EXIF_DATE_CHARS && isdigit((unsigned char)cstr[0]) && cstr[length-1]==0) { - char cstr_writable[JPEG_EXIF_DATE_CHARS]; - strcpy(cstr_writable, cstr); - try { - tm tmIn=cstr_to_time_t(cstr_writable, 0); + tm tmIn=cstr_to_time_t(pa_strdup(cstr), 0); return new VDate(tmIn); } catch(...) { /*ignore bad date times*/ } @@ -485,14 +453,14 @@ static Value* parse_IFD_entry_formatted_ HashStringValue& hash=result->hash(); for(uint i=0; ibyte_align_identifier[0]=='M'; // [M]otorola vs [I]ntel uint first_IFD_offset=endian_to_uint(is_big, head->first_IFD_offset); - reader.seek(tiff_base+first_IFD_offset, SEEK_SET); + reader.seek(tiff_base+first_IFD_offset); VHash* vhash=new VHash; @@ -631,7 +599,7 @@ static void measure_jpeg(const String& o throw Exception(IMAGE_FORMAT, &origin_string, "not JPEG file - wrong signature"); while(true) { - uint segment_base=reader.tell()+2/*marker,code*/; + uint64_t segment_base=reader.tell()+2/*marker,code*/; if(reader.read(buf, sizeof(JPG_Segment_head))width); @@ -685,12 +653,14 @@ static void measure_jpeg(const String& o return; }; - reader.seek(segment_base + segment_length, SEEK_SET); + reader.seek(segment_base + segment_length); } throw Exception(IMAGE_FORMAT, &origin_string, "broken JPEG file - size frame not found"); } +///*********************************************** TIFF + static bool parse_tiff_IFD(bool is_big, Measure_reader& reader, Measure_info &info) { const char* buf; if(reader.read(buf, sizeof(JPG_Exif_IFD_begin))components_count) != 1) return false; uint value = (entry_format == 3) ? endian_to_ushort(is_big, entry->value_or_offset_to_it) : endian_to_uint(is_big, entry->value_or_offset_to_it); - (entry_tag == 256) ? info.width=value : info.height=value; + (entry_tag == 256) ? info.width=(short)value : info.height=(short)value; if(info.width && info.height) return true; } @@ -736,11 +706,24 @@ static void measure_tiff(const String& o if(endian_to_ushort(is_big, head->signature) != 42) throw Exception(IMAGE_FORMAT, &origin_string, "not TIFF file - wrong signature"); - reader.seek(endian_to_uint(is_big, head->first_IFD_offset), SEEK_SET); + reader.seek(endian_to_uint(is_big, head->first_IFD_offset)); if(!parse_tiff_IFD(is_big, reader, info)) throw Exception(IMAGE_FORMAT, &origin_string, "broken TIFF file - size field entry not found"); } + +///*********************************************** GIF + +struct GIF_Header { + char signature[3]; // 'GIF' + char version[3]; + uchar width[2]; + uchar height[2]; + char dif; + char fonColor; + char nulls; +}; + static void measure_gif(const String& origin_string, Measure_reader& reader, ushort& width, ushort& height) { const char* buf; @@ -756,6 +739,18 @@ static void measure_gif(const String& or height=endian_to_ushort(false, head->height); } + +///*********************************************** PNG + +struct PNG_Header { + char dummy[12]; + char signature[4]; //< must be "IHDR" + uchar high_width[2]; //< image width high bytes [we ignore for now] + uchar width[2]; //< image width low bytes + uchar high_height[2]; //< image height high bytes [we ignore for now] + uchar height[4]; //< image height +}; + static void measure_png(const String& origin_string, Measure_reader& reader, ushort& width, ushort& height) { const char* buf; @@ -771,6 +766,21 @@ static void measure_png(const String& or height=endian_to_ushort(true, head->height); } + +///*********************************************** BMP + +struct BMP_Header { + char signature[2]; // 'BM' + uchar file_size[4]; + uchar reserved[4]; + uchar bitmap_offset[4]; + uchar header_size[4]; + uchar width[2]; + uchar high_width[2]; //< image width high bytes [we ignore for now] + uchar height[2]; + uchar high_height[2]; //< image height high bytes [we ignore for now] +}; + static void measure_bmp(const String& origin_string, Measure_reader& reader, ushort& width, ushort& height) { const char* buf; @@ -782,15 +792,164 @@ static void measure_bmp(const String& or if(strncmp(head->signature, "BM", 2)!=0) throw Exception(IMAGE_FORMAT, &origin_string, "not BMP file - wrong signature"); - reader.seek(0, SEEK_END); - if((uint)reader.tell() != endian_to_uint(false, head->file_size)) + if((uint)reader.length() != endian_to_uint(false, head->file_size)) throw Exception(IMAGE_FORMAT, &origin_string, "not BMP file - length header and file size do not match"); width=endian_to_ushort(false, head->width); height=endian_to_ushort(false, head->height); } -// measure center + +///*********************************************** WEBP + +struct WEBP_Header { + char signature_riff[4]; // 'RIFF' + uchar file_size[4]; + char signature[4]; // 'WEBP' + char format[4]; // 'VP8 ' or 'VP8L' or 'VP8X' +}; + +struct WEBP_VP8_Chunk { + uchar size[4]; + char tag[3]; + uchar signature[3]; // 0x9D 0x01 0x2A + uchar width[2]; // 14 bits each + uchar height[2]; // 14 bits each +}; + +struct WEBP_VP8L_Chunk { + uchar size[4]; + char signature; // 0x2F + uchar width_height[4]; // 14 bits each +}; + +struct WEBP_X_Chunk { + uchar size[4]; + char reserved[4]; + uchar width[3]; + uchar height[3]; +}; + +static void measure_webp(const String& origin_string, Measure_reader& reader, ushort& width, ushort& height) { + const char* buf; + + if(reader.read(buf, sizeof(WEBP_Header))signature_riff, "RIFF", 4)!=0 || strncmp(head->signature, "WEBP", 4)!=0) + throw Exception(IMAGE_FORMAT, &origin_string, "not WEBP file - wrong signature"); + + if(strncmp(head->format, "VP8 ", 4)==0){ + if(reader.read(buf, sizeof(WEBP_VP8_Chunk))signature[0] != 0x9D || chunk->signature[1] != 0x01 || chunk->signature[2] != 0x2A) + throw Exception(IMAGE_FORMAT, &origin_string, "broken WEBP file - wrong VP8 chunk signature"); + + width=endian_to_ushort(false, chunk->width) & 0x3FFF; + height=endian_to_ushort(false, chunk->height) & 0x3FFF; + } else if(strncmp(head->format, "VP8L", 4)==0){ + if(reader.read(buf, sizeof(WEBP_VP8L_Chunk))signature != 0x2F) + throw Exception(IMAGE_FORMAT, &origin_string, "broken WEBP file - wrong VP8L chunk signature"); + + uint wh=endian_to_uint(false, chunk->width_height); + width=(wh & 0x3FFF) + 1; + height=((wh >> 14) & 0x3FFF) + 1; + } else if (strncmp(head->format, "VP8X", 4)==0){ + if(reader.read(buf, sizeof(WEBP_X_Chunk))width) + 1; // we ignore third byte to simplify code + height=endian_to_ushort(false, chunk->height) + 1; // we ignore third byte to simplify code + } else throw Exception(IMAGE_FORMAT, &origin_string, "broken WEBP file - invalid chunk signature"); +} + + +///*********************************************** MP4 + +struct MP4_Header { + uchar size[4]; + char signature[4]; // 'ftyp' in first chunk +}; + +struct MP4_ExtSize { + uchar high[4]; + uchar low[4]; +}; + +struct MP4_Tkhd { + uchar width[4]; + uchar height[4]; +}; + +static bool measure_mp4(const String& origin_string, Measure_reader& reader, ushort& width, ushort& height, uint64_t anext, const char* lastTkhd=NULL) { + for(bool first=anext==0;;){ + const char* buf; + uint64_t next=reader.tell(); + + if(reader.read(buf, sizeof(MP4_Header))size); + + if(size==1){ + if(reader.read(buf, sizeof(MP4_ExtSize))high) << 32) + endian_to_uint(true, ext_size->low); + } + next+=size; + + if(first){ + if(strncmp(head->signature, "ftyp", 4)!=0) + throw Exception(IMAGE_FORMAT, &origin_string, "not MP4 file - wrong signature"); + first=false; + anext=reader.length(); // to avoid reading beyond EOF + } else if(strncmp(head->signature, "moov", 4)==0 || strncmp(head->signature, "mdia", 4)==0 || strncmp(head->signature, "trak", 4)==0) { + if(measure_mp4(origin_string, reader, width, height, next, lastTkhd)) + return true; + } else if(strncmp(head->signature, "tkhd", 4)==0) { + if(size>8){ + reader.seek(next-8); + if(reader.read(lastTkhd, sizeof(MP4_Tkhd))signature, "hdlr", 4)==0) { + if(size>12){ + const char* hdlr; + if(reader.read(hdlr, 12)<12) + throw Exception(IMAGE_FORMAT, &origin_string, "broken MP4 file - bad hdlr chunk"); + if(lastTkhd && strncmp(hdlr+8, "vide", 4)==0) { + MP4_Tkhd *tkhd=(MP4_Tkhd *)lastTkhd; + width=endian_to_ushort(true, tkhd->width); + height=endian_to_ushort(true, tkhd->height); + return true; + } + } + } + if(anext && next>=anext) + break; + reader.seek(next); + } + return false; +} + +static void measure_mp4(const String& origin_string, Measure_reader& reader, ushort& width, ushort& height) { + if(!measure_mp4(origin_string, reader, width, height, 0)) + throw Exception(IMAGE_FORMAT, &origin_string, "unsupported MP4 file - size not found"); +} + +///*********************************************** measure center static void measure(const String& file_name, Measure_reader& reader, Measure_info &info) { const char* file_name_cstr=file_name.taint_cstr(String::L_FILE_SPEC); @@ -804,12 +963,19 @@ static void measure(const String& file_n measure_png(file_name, reader, info.width, info.height); else if(strcasecmp(cext, "BMP")==0) measure_bmp(file_name, reader, info.width, info.height); + else if(strcasecmp(cext, "WEBP")==0) + measure_webp(file_name, reader, info.width, info.height); else if(strcasecmp(cext, "TIF")==0 || strcasecmp(cext, "TIFF")==0) measure_tiff(file_name, reader, info); + else if(strcasecmp(cext, "MP4")==0 || strcasecmp(cext, "MOV")==0) + if(info.video) + measure_mp4(file_name, reader, info.width, info.height); + else + throw Exception(IMAGE_FORMAT, &file_name, "handling disabled for file name extension '%s'", cext); else - throw Exception(IMAGE_FORMAT, &file_name, "unhandled image file name extension '%s'", cext); + throw Exception(IMAGE_FORMAT, &file_name, "unhandled file name extension '%s'", cext); } else - throw Exception(IMAGE_FORMAT, &file_name, "can not determine image type - no file name extension"); + throw Exception(IMAGE_FORMAT, &file_name, "cannot determine file type - no file name extension"); } // methods @@ -824,7 +990,7 @@ static void _measure(Request& r, MethodP Value* exif=0; Value* xmp=0; - Measure_info info={ 0, 0, 0, 0, &pa_UTF8_charset }; + Measure_info info={ 0, 0, 0, 0, &pa_UTF8_charset, false }; if(params.count()>1) if(HashStringValue* options=params.as_hash(1, "methods options")) { @@ -846,6 +1012,10 @@ static void _measure(Request& r, MethodP info.xmp_charset=&pa_charsets.get(value->as_string()); valid_options++; } + if(key == "video") { + info.video=r.process(*value).as_bool(); + valid_options++; + } } if(valid_options!=options->count()) throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION); @@ -854,9 +1024,9 @@ static void _measure(Request& r, MethodP const String* file_name; if(file_name=data.get_string()) { - file_read_action_under_lock(r.absolute(*file_name), "measure", file_measure_action, &info); + file_read_action_under_lock(r.full_disk_path(*file_name), "measure", file_measure_action, &info); } else { - VFile* vfile=data.as_vfile(String::L_AS_IS); + VFile* vfile=data.as_vfile(); file_name=&vfile->fields().get(name_name)->as_string(); Measure_buf_reader reader(vfile->value_ptr(), vfile->value_size(), *file_name); measure(*file_name, reader, info); @@ -911,7 +1081,7 @@ static void _html(Request& r, MethodPara /// @test wrap FILE to auto-object static gdImage* load(Request& r, const String& file_name){ - const char* file_name_cstr=r.absolute(file_name).taint_cstr(String::L_FILE_SPEC); + const char* file_name_cstr=r.full_disk_path(file_name).taint_cstr(String::L_FILE_SPEC); if(FILE *f=pa_fopen(file_name_cstr, "rb")) { gdImage* image=new gdImage; bool ok=image->CreateFromGif(f); @@ -920,13 +1090,13 @@ static gdImage* load(Request& r, const S throw Exception(IMAGE_FORMAT, &file_name, "is not in GIF format"); return image; } else { - throw Exception("file.missing", 0, "can not open '%s'", file_name_cstr); + throw Exception("file.missing", 0, "cannot open '%s'", file_name_cstr); } } static void _load(Request& r, MethodParams& params) { - const String& file_name=params.as_string(0, FILE_NAME_MUST_NOT_BE_CODE); + const String& file_name=params.as_file_name(0); gdImage* image=load(r, file_name); GET_SELF(r, VImage).set(&file_name, image->SX(), image->SY(), image); @@ -1025,11 +1195,12 @@ static void _replace(Request& r, MethodP gdImage::Point* all_p=0; size_t count=0; if(params.count() == 3){ - Table* table=params.as_table(2, "coordinates"); - count=table->count(); - all_p=new(PointerFreeGC) gdImage::Point[count]; - gdImage::Point* add_p=all_p; - table->for_each(add_point, &add_p); + if(Table* table=params.as_table(2, "coordinates")){ + count=table->count(); + all_p=new(PointerFreeGC) gdImage::Point[count]; + gdImage::Point* add_p=all_p; + table->for_each(add_point, &add_p); + } } else { int max_x=image.SX()-1; int max_y=image.SY()-1; @@ -1051,38 +1222,34 @@ static void _replace(Request& r, MethodP static void _polyline(Request& r, MethodParams& params) { gdImage& image=GET_SELF(r, VImage).image(); - Table* table=params.as_table(1, "coordinates"); - - gdImage::Point* all_p=new(PointerFreeGC) gdImage::Point[table->count()]; - gdImage::Point *add_p=all_p; - table->for_each(add_point, &add_p); - image.Polygon(all_p, table->count(), - image.Color(params.as_int(0, "color must be int", r)), - false/*not closed*/); + if(Table* table=params.as_table(1, "coordinates")){ + gdImage::Point* all_p=new(PointerFreeGC) gdImage::Point[table->count()]; + gdImage::Point *add_p=all_p; + table->for_each(add_point, &add_p); + image.Polygon(all_p, table->count(), image.Color(params.as_int(0, "color must be int", r)), false/*not closed*/); + } } static void _polygon(Request& r, MethodParams& params) { gdImage& image=GET_SELF(r, VImage).image(); - Table* table=(Table*)params.as_table(1, "coordinates"); - - gdImage::Point* all_p=new(PointerFreeGC) gdImage::Point[table->count()]; - gdImage::Point *add_p=all_p; - table->for_each(add_point, &add_p); - image.Polygon(all_p, table->count(), - image.Color(params.as_int(0, "color must be int", r))); + if(Table* table=(Table*)params.as_table(1, "coordinates")){ + gdImage::Point* all_p=new(PointerFreeGC) gdImage::Point[table->count()]; + gdImage::Point *add_p=all_p; + table->for_each(add_point, &add_p); + image.Polygon(all_p, table->count(), image.Color(params.as_int(0, "color must be int", r))); + } } static void _polybar(Request& r, MethodParams& params) { gdImage& image=GET_SELF(r, VImage).image(); - Table* table=(Table*)params.as_table(1, "coordinates"); - - gdImage::Point* all_p=new(PointerFreeGC) gdImage::Point[table->count()]; - gdImage::Point *add_p=all_p; - table->for_each(add_point, &add_p); - image.FilledPolygon(all_p, table->count(), - image.Color(params.as_int(0, "color must be int", r))); + if(Table* table=(Table*)params.as_table(1, "coordinates")){ + gdImage::Point* all_p=new(PointerFreeGC) gdImage::Point[table->count()]; + gdImage::Point *add_p=all_p; + table->for_each(add_point, &add_p); + image.FilledPolygon(all_p, table->count(), image.Color(params.as_int(0, "color must be int", r))); + } } // font @@ -1185,7 +1352,7 @@ static void _font(Request& r, MethodPara if(!alphabet_length) throw Exception(PARSER_RUNTIME, 0, "alphabet must not be empty"); - gdImage* image=load(r, params.as_string(1, FILE_NAME_MUST_NOT_BE_CODE)); + gdImage* image=load(r, params.as_file_name(1)); int spacebar_width=image->SX(); int monospace_width=0; // proportional @@ -1288,8 +1455,8 @@ static void _circle(Request& r, MethodPa gdImage& as_image(MethodParams& params, int index, const char* msg) { Value& value=params.as_no_junction(index, msg); - if(Value* vimage=value.as(VIMAGE_TYPE)) { - return static_cast(vimage)->image(); + if(VImage* vimage=dynamic_cast(&value)) { + return vimage->image(); } else throw Exception(PARSER_RUNTIME, 0, msg); }