OpenSystemsDevelopment / qpdf

Commit fa3051d977ab21da7bde9d9613dca600c83123a5

Authored by Jay Berkenbilt
1 parent 2008d037

Implement --json-keys

Inline Side-by-side
Showing 1 changed file with 110 additions and 59 deletions
qpdf/qpdf.cc
  View file @fa3051d
... ... @@ -194,6 +194,7 @@ struct Options
194 194 bool show_pages;
195 195 bool show_page_images;
196 196 bool json;
  197 + std::set<std::string> json_keys;
197 198 bool check;
198 199 std::vector<PageSpec> page_specs;
199 200 std::map<std::string, RotationSpec> rotations;
... ... @@ -243,7 +244,7 @@ ProgressReporter::reportProgress(int percentage)
243 244 << percentage << "%" << std::endl;
244 245 }
245 246  
246   -static JSON json_schema()
  247 +static JSON json_schema(std::set<std::string>* keys = 0)
247 248 {
248 249 // Style: use all lower-case keys with no dashes or underscores.
249 250 // Choose array or dictionary based on indexing. For example, we
... ... @@ -270,60 +271,76 @@ static JSON json_schema()
270 271 j_params.addDictionaryMember(
271 272 "decodelevel", JSON::makeString(
272 273 "decode level used to determine stream filterability"));
273   - schema.addDictionaryMember(
274   - "objects", JSON::makeString(
275   - "dictionary of original objects; keys are 'trailer' or 'n n R'"));
276   - JSON page = schema.addDictionaryMember("pages", JSON::makeArray()).
277   - addArrayElement(JSON::makeDictionary());
278   - page.addDictionaryMember(
279   - "object",
280   - JSON::makeString("reference to original page object"));
281   - JSON image = page.addDictionaryMember("images", JSON::makeArray()).
282   - addArrayElement(JSON::makeDictionary());
283   - image.addDictionaryMember(
284   - "object",
285   - JSON::makeString("reference to image stream"));
286   - image.addDictionaryMember(
287   - "width",
288   - JSON::makeString("image width"));
289   - image.addDictionaryMember(
290   - "height",
291   - JSON::makeString("image height"));
292   - image.addDictionaryMember("filter", JSON::makeArray()).
293   - addArrayElement(
294   - JSON::makeString("filters applied to image data"));
295   - image.addDictionaryMember("decodeparms", JSON::makeArray()).
296   - addArrayElement(
297   - JSON::makeString("decode parameters for image data"));
298   - image.addDictionaryMember(
299   - "filterable",
300   - JSON::makeString("whether image data can be decoded"
301   - " using the decode level qpdf was invoked with"));
302   - page.addDictionaryMember("contents", JSON::makeArray()).
303   - addArrayElement(
304   - JSON::makeString("reference to each content stream"));
305   - page.addDictionaryMember(
306   - "label",
307   - JSON::makeString("page label dictionary, or null if none"));
308   - JSON labels = schema.addDictionaryMember("pagelabels", JSON::makeArray()).
309   - addArrayElement(JSON::makeDictionary());
310   - labels.addDictionaryMember(
311   - "index",
312   - JSON::makeString("starting page position starting from zero"));
313   - labels.addDictionaryMember(
314   - "label",
315   - JSON::makeString("page label dictionary"));
316   - JSON outline = page.addDictionaryMember("outlines", JSON::makeArray()).
317   - addArrayElement(JSON::makeDictionary());
318   - outline.addDictionaryMember(
319   - "object",
320   - JSON::makeString("reference to outline that targets this page"));
321   - outline.addDictionaryMember(
322   - "title",
323   - JSON::makeString("outline title"));
324   - outline.addDictionaryMember(
325   - "dest",
326   - JSON::makeString("outline destination dictionary"));
  274 +
  275 + bool all_keys = ((keys == 0) || keys->empty());
  276 +
  277 + // The list of selectable top-level keys id duplicated in three
  278 + // places: json_schema, do_json, and initOptionTable.
  279 + if (all_keys || keys->count("objects"))
  280 + {
  281 + schema.addDictionaryMember(
  282 + "objects", JSON::makeString(
  283 + "dictionary of original objects;"
  284 + " keys are 'trailer' or 'n n R'"));
  285 + }
  286 + if (all_keys || keys->count("pages"))
  287 + {
  288 + JSON page = schema.addDictionaryMember("pages", JSON::makeArray()).
  289 + addArrayElement(JSON::makeDictionary());
  290 + page.addDictionaryMember(
  291 + "object",
  292 + JSON::makeString("reference to original page object"));
  293 + JSON image = page.addDictionaryMember("images", JSON::makeArray()).
  294 + addArrayElement(JSON::makeDictionary());
  295 + image.addDictionaryMember(
  296 + "object",
  297 + JSON::makeString("reference to image stream"));
  298 + image.addDictionaryMember(
  299 + "width",
  300 + JSON::makeString("image width"));
  301 + image.addDictionaryMember(
  302 + "height",
  303 + JSON::makeString("image height"));
  304 + image.addDictionaryMember("filter", JSON::makeArray()).
  305 + addArrayElement(
  306 + JSON::makeString("filters applied to image data"));
  307 + image.addDictionaryMember("decodeparms", JSON::makeArray()).
  308 + addArrayElement(
  309 + JSON::makeString("decode parameters for image data"));
  310 + image.addDictionaryMember(
  311 + "filterable",
  312 + JSON::makeString("whether image data can be decoded"
  313 + " using the decode level qpdf was invoked with"));
  314 + page.addDictionaryMember("contents", JSON::makeArray()).
  315 + addArrayElement(
  316 + JSON::makeString("reference to each content stream"));
  317 + page.addDictionaryMember(
  318 + "label",
  319 + JSON::makeString("page label dictionary, or null if none"));
  320 + JSON outline = page.addDictionaryMember("outlines", JSON::makeArray()).
  321 + addArrayElement(JSON::makeDictionary());
  322 + outline.addDictionaryMember(
  323 + "object",
  324 + JSON::makeString("reference to outline that targets this page"));
  325 + outline.addDictionaryMember(
  326 + "title",
  327 + JSON::makeString("outline title"));
  328 + outline.addDictionaryMember(
  329 + "dest",
  330 + JSON::makeString("outline destination dictionary"));
  331 + }
  332 + if (all_keys || keys->count("pagelabels"))
  333 + {
  334 + JSON labels = schema.addDictionaryMember(
  335 + "pagelabels", JSON::makeArray()).
  336 + addArrayElement(JSON::makeDictionary());
  337 + labels.addDictionaryMember(
  338 + "index",
  339 + JSON::makeString("starting page position starting from zero"));
  340 + labels.addDictionaryMember(
  341 + "label",
  342 + JSON::makeString("page label dictionary"));
  343 + }
327 344 return schema;
328 345 }
329 346  
... ... @@ -419,6 +436,7 @@ class ArgParser
419 436 void argShowPages();
420 437 void argWithImages();
421 438 void argJson();
  439 + void argJsonKey(char* parameter);
422 440 void argCheck();
423 441 void arg40Print(char* parameter);
424 442 void arg40Modify(char* parameter);
... ... @@ -615,6 +633,11 @@ ArgParser::initOptionTable()
615 633 (*t)["show-pages"] = oe_bare(&ArgParser::argShowPages);
616 634 (*t)["with-images"] = oe_bare(&ArgParser::argWithImages);
617 635 (*t)["json"] = oe_bare(&ArgParser::argJson);
  636 + // The list of selectable top-level keys id duplicated in three
  637 + // places: json_schema, do_json, and initOptionTable.
  638 + char const* jsonKeyChoices[] = {"objects", "pages", "pagelabels", 0};
  639 + (*t)["json-key"] = oe_requiredChoices(
  640 + &ArgParser::argJsonKey, jsonKeyChoices);
618 641 (*t)["check"] = oe_bare(&ArgParser::argCheck);
619 642  
620 643 t = &this->encrypt40_option_table;
... ... @@ -1178,6 +1201,12 @@ ArgParser::argJson()
1178 1201 }
1179 1202  
1180 1203 void
  1204 +ArgParser::argJsonKey(char* parameter)
  1205 +{
  1206 + o.json_keys.insert(parameter);
  1207 +}
  1208 +
  1209 +void
1181 1210 ArgParser::argCheck()
1182 1211 {
1183 1212 o.check = true;
... ... @@ -1659,6 +1688,16 @@ automated test suites for software that uses the qpdf library.\n\
1659 1688 --show-pages shows the object/generation number for each page\n\
1660 1689 --with-images also shows the object IDs for images on each page\n\
1661 1690 --check check file structure + encryption, linearization\n\
  1691 +--json generate a json representation of the file\n\
  1692 +--json-help describe the format of the json representation\n\
  1693 +--json-key=key repeatable; prune json structure to include only\n\
  1694 + specified keys\n\
  1695 +\n\
  1696 +The json representation generated by qpdf is designed to facilitate\n\
  1697 +processing of qpdf from other programming languages that have a hard\n\
  1698 +time calling C++ APIs. Run qpdf --json-help for details on the format.\n\
  1699 +The manual has more in-depth information about the json representation\n\
  1700 +and certain compatibility guarantees that qpdf provides.\n\
1662 1701 \n\
1663 1702 The --raw-stream-data and --filtered-stream-data options are ignored\n\
1664 1703 unless --show-object is given. Either of these options will cause the\n\
... ... @@ -2704,13 +2743,25 @@ static void do_json(QPDF&amp; pdf, Options&amp; o)
2704 2743 j_params.addDictionaryMember(
2705 2744 "decodelevel", JSON::makeString(decode_level_str));
2706 2745  
2707   - do_json_objects(pdf, o, j);
2708   - do_json_pages(pdf, o, j);
2709   - do_json_page_labels(pdf, o, j);
  2746 + bool all_keys = o.json_keys.empty();
  2747 + // The list of selectable top-level keys id duplicated in three
  2748 + // places: json_schema, do_json, and initOptionTable.
  2749 + if (all_keys || o.json_keys.count("objects"))
  2750 + {
  2751 + do_json_objects(pdf, o, j);
  2752 + }
  2753 + if (all_keys || o.json_keys.count("pages"))
  2754 + {
  2755 + do_json_pages(pdf, o, j);
  2756 + }
  2757 + if (all_keys || o.json_keys.count("pagelabels"))
  2758 + {
  2759 + do_json_page_labels(pdf, o, j);
  2760 + }
2710 2761  
2711 2762 // Check against schema
2712 2763  
2713   - JSON schema = json_schema();
  2764 + JSON schema = json_schema(&o.json_keys);
2714 2765 std::list<std::string> errors;
2715 2766 if (! j.checkSchema(schema, errors))
2716 2767 {
... ...