--- parser3/src/classes/image.C 2017/05/16 14:42:07 1.162 +++ parser3/src/classes/image.C 2020/12/01 18:53:01 1.172 @@ -24,8 +24,9 @@ #include "pa_vimage.h" #include "pa_vdate.h" #include "pa_table.h" +#include "pa_charsets.h" -volatile const char * IDENT_IMAGE_C="$Id: image.C,v 1.162 2017/05/16 14:42:07 moko Exp $"; +volatile const char * IDENT_IMAGE_C="$Id: image.C,v 1.172 2020/12/01 18:53:01 moko Exp $"; // defines @@ -227,14 +228,14 @@ public: } } exif_gps_tag_value2name; -#undef EXIF_TAG -#ifndef DOXYGEN +///*********************************************** support functions + 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(off_t value, int whence)=0; + virtual off_t tell()=0; }; class Measure_file_reader: public Measure_reader { @@ -257,12 +258,12 @@ public: return read_size; } - override void seek(long value, int whence) { + override void seek(off_t 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 long tell() { return lseek(f, 0, SEEK_CUR); } + override off_t tell() { return lseek(f, 0, SEEK_CUR); } }; @@ -285,11 +286,12 @@ public: return to_read; } - override void seek(long value, int whence) { + override void seek(off_t 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 @@ -301,34 +303,39 @@ public: offset=new_offset; } - override long tell() { return offset; } + override off_t tell() { return offset; } }; -#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; }; -/// 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; -}; -/// JPEG record head +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 + struct JPG_Segment_head { uchar marker; uchar code; @@ -336,16 +343,16 @@ struct JPG_Segment_head { }; /// JPEG frame header struct JPG_Size_segment_body { - char data; //< data precision of bits/sample - uchar height[2]; //< image height - uchar width[2]; //< image width - char numComponents; //< number of color components + char data; //< data precision of bits/sample + uchar height[2]; //< image height + uchar width[2]; //< image width + char numComponents; //< number of color components }; /// JPEG Exif TIFF Header struct JPG_Exif_TIFF_header { - uchar byte_align_identifier[2]; - char dummy[2]; // always 000A [or 0A00] + char byte_align_identifier[2]; + uchar signature[2]; // always 000A [or 0A00] uchar first_IFD_offset[4]; // Usually the first IFD starts immediately next to TIFF header, so this offset has value '0x00000008'. }; @@ -368,51 +375,6 @@ struct JPG_Exif_IFD_entry { #define JPEG_EXIF_DATE_CHARS 20 - -#ifndef DOXYGEN -struct Measure_info { - ushort width; - ushort height; - Value** exif; - Value** xmp; -}; -#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 void measure_gif(const String& origin_string, Measure_reader& reader, ushort& width, ushort& height) { - - const char* buf; - const size_t head_size=sizeof(GIF_Header); - if(reader.read(buf, head_size)signature, "GIF", 3)!=0) - throw Exception(IMAGE_FORMAT, &origin_string, "not GIF file - wrong signature"); - - width=endian_to_ushort(false, head->width); - height=endian_to_ushort(false, head->height); -} - 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 @@ -528,7 +490,7 @@ static Value* parse_IFD_entry_value(bool if(value_size<=4) result=parse_IFD_entry_formatted_value(is_big, format, component_size, components_count, entry.value_or_offset_to_it); else { - long remembered=reader.tell(); + off_t remembered=reader.tell(); { reader.seek(tiff_base+endian_to_uint(is_big, entry.value_or_offset_to_it), SEEK_SET); const char* value; @@ -548,7 +510,7 @@ static void parse_IFD_entry(HashStringVa ushort tag=endian_to_ushort(is_big, entry.tag); if(tag==JPG_IFD_TAG_EXIF_OFFSET || tag==JPG_IFD_TAG_EXIF_GPS_OFFSET){ - long remembered=reader.tell(); + off_t remembered=reader.tell(); { reader.seek(tiff_base+endian_to_uint(is_big, entry.value_or_offset_to_it), SEEK_SET); parse_IFD(hash, is_big, reader, tiff_base, (tag==JPG_IFD_TAG_EXIF_GPS_OFFSET)?true:gps); @@ -602,11 +564,13 @@ static Value* parse_exif(Measure_reader& return vhash; } -static Value* parse_xmp(Measure_reader& reader, ushort xmp_length) { +static Value* parse_xmp(Measure_reader& reader, ushort xmp_length, Measure_info &info) { const char* buf; if(reader.read(buf, xmp_length)directory_entry_count); + for(int i=0; itag); + + if(entry_tag == 256 || entry_tag == 257){ + ushort entry_format=endian_to_ushort(is_big, entry->format); + if(entry_format != 3 && entry_format != 4 || endian_to_uint(is_big, entry->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; + if(info.width && info.height) + return true; + } + } + + return false; + // then goes: LLLLLLLL Offset to next IFD [not going there] +} + +static void measure_tiff(const String& origin_string, Measure_reader& reader, Measure_info &info) { + const char* buf; + + if(reader.read(buf, sizeof(JPG_Exif_TIFF_header))byte_align_identifier, "II", 2)!=0 && strncmp(head->byte_align_identifier, "MM", 2)!=0) + throw Exception(IMAGE_FORMAT, &origin_string, "not TIFF file - wrong signature"); + + bool is_big=head->byte_align_identifier[0]=='M'; // [M]otorola vs [I]ntel + + 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); + 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; + const size_t head_size=sizeof(GIF_Header); + if(reader.read(buf, head_size)signature, "GIF", 3)!=0) + throw Exception(IMAGE_FORMAT, &origin_string, "not GIF file - wrong signature"); + + width=endian_to_ushort(false, head->width); + 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; @@ -703,7 +759,178 @@ static void measure_png(const String& or height=endian_to_ushort(true, head->height); } -// measure center + +///*********************************************** 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; + const size_t head_size=sizeof(BMP_Header); + if(reader.read(buf, head_size)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)) + 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); +} + + +///*********************************************** 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 +}; + +static bool measure_mp4(const String& origin_string, Measure_reader& reader, ushort& width, ushort& height, off_t anext, const char* lastTkhd=NULL) { + for(bool first=anext==0;;){ + const char* buf; + const size_t head_size=sizeof(MP4_Header); + if(reader.read(buf, head_size)size); + off_t next=reader.tell() + size - head_size; + +// printf("%d processing chunk size %d signature '%c%c%c%c %p'\n", anext, size, head->signature[0], head->signature[1], head->signature[2], head->signature[3], lastTkhd); + + if(first){ + if(strncmp(head->signature, "ftyp", 4)!=0) + throw Exception(IMAGE_FORMAT, &origin_string, "not MP4 file - wrong signature"); + first=false; + reader.seek(0, SEEK_END); + anext=reader.tell(); // 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, SEEK_SET); + if(reader.read(lastTkhd, 8)<8) + throw Exception(IMAGE_FORMAT, &origin_string, "broken MP4 file - bad tkhd chunk"); + } + } else if (strncmp(head->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(strncmp(hdlr+8, "vide", 4)==0) +// printf("vide found\n"); + if( lastTkhd && strncmp(hdlr+8, "vide", 4)==0) { + width=endian_to_ushort(true, (const unsigned char*)(lastTkhd)); + height=endian_to_ushort(true, (const unsigned char*)(lastTkhd+4)); +// printf("wh %d %d\n",width, height); + return true; + } + } + } + if(anext && next>=anext) + break; + reader.seek(next, SEEK_SET); + } + 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); @@ -711,10 +938,18 @@ static void measure(const String& file_n cext++; if(strcasecmp(cext, "GIF")==0) measure_gif(file_name, reader, info.width, info.height); - else if(strcasecmp(cext, "JPG")==0 || strcasecmp(cext, "JPEG")==0) + else if(strcasecmp(cext, "JPG")==0 || strcasecmp(cext, "JPEG")==0) measure_jpeg(file_name, reader, info); else if(strcasecmp(cext, "PNG")==0) 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) + measure_mp4(file_name, reader, info.width, info.height); else throw Exception(IMAGE_FORMAT, &file_name, "unhandled image file name extension '%s'", cext); } else @@ -733,12 +968,37 @@ static void _measure(Request& r, MethodP Value* exif=0; Value* xmp=0; - Measure_info info={ 0, 0, &exif, &xmp }; + Measure_info info={ 0, 0, 0, 0, &pa_UTF8_charset }; + + if(params.count()>1) + if(HashStringValue* options=params.as_hash(1, "methods options")) { + int valid_options=0; + for(HashStringValue::Iterator i(*options); i; i.next() ){ + String::Body key=i.key(); + Value* value=i.value(); + if(key == "exif") { + if(r.process(*value).as_bool()) + info.exif=&exif; + valid_options++; + } + if(key == "xmp") { + if(r.process(*value).as_bool()) + info.xmp=&xmp; + valid_options++; + } + if(key == "xmp-charset") { + info.xmp_charset=&pa_charsets.get(value->as_string()); + valid_options++; + } + } + if(valid_options!=options->count()) + throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION); + } 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); file_name=&vfile->fields().get(name_name)->as_string(); @@ -795,7 +1055,7 @@ 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); @@ -1219,7 +1479,8 @@ static void _pixel(Request& r, MethodPar MImage::MImage(): Methoded("image") { // ^image:measure[DATA] - add_native_method("measure", Method::CT_DYNAMIC, _measure, 1, 1); + // ^image:measure[DATA; $.exif(false) $.xmp(false) $.xmp-charset[UTF-8] ] + add_native_method("measure", Method::CT_DYNAMIC, _measure, 1, 2); // ^image.html[] // ^image.html[hash]