OpenSystemsDevelopment / qpdf

Commit ae90619f9fe8ac6103faa105b7750acb41d199c8

Authored by m-holger
1 parent cc9facdb

Refactor `pushInheritedAttributesToPage`: move implementation to `QPDF_pages.cc`… 

… to improve code organization.
Inline Side-by-side
Showing 2 changed files with 131 additions and 131 deletions
libqpdf/QPDF_optimization.cc
  View file @ae90619
... ... @@ -135,137 +135,6 @@ Lin::optimize_internal(
135 135 }
136 136  
137 137 void
138   -QPDF::pushInheritedAttributesToPage()
139   -{
140   - // Public API should not have access to allow_changes.
141   - m->pages.pushInheritedAttributesToPage(true, false);
142   -}
143   -
144   -void
145   -Pages::pushInheritedAttributesToPage(bool allow_changes, bool warn_skipped_keys)
146   -{
147   - // Traverse pages tree pushing all inherited resources down to the page level.
148   -
149   - // The record of whether we've done this is cleared by updateAllPagesCache(). If we're warning
150   - // for skipped keys, re-traverse unconditionally.
151   - if (m->pushed_inherited_attributes_to_pages && (!warn_skipped_keys)) {
152   - return;
153   - }
154   -
155   - // Calling getAllPages() resolves any duplicated page objects, repairs broken nodes, and detects
156   - // loops, so we don't have to do those activities here.
157   - qpdf.getAllPages();
158   -
159   - // key_ancestors is a mapping of page attribute keys to a stack of Pages nodes that contain
160   - // values for them.
161   - std::map<std::string, std::vector<QPDFObjectHandle>> key_ancestors;
162   - pushInheritedAttributesToPageInternal(
163   - m->trailer.getKey("/Root").getKey("/Pages"),
164   - key_ancestors,
165   - allow_changes,
166   - warn_skipped_keys);
167   - if (!key_ancestors.empty()) {
168   - throw std::logic_error(
169   - "key_ancestors not empty after pushing inherited attributes to pages");
170   - }
171   - m->pushed_inherited_attributes_to_pages = true;
172   - m->ever_pushed_inherited_attributes_to_pages = true;
173   -}
174   -
175   -void
176   -Pages ::pushInheritedAttributesToPageInternal(
177   - QPDFObjectHandle cur_pages,
178   - std::map<std::string, std::vector<QPDFObjectHandle>>& key_ancestors,
179   - bool allow_changes,
180   - bool warn_skipped_keys)
181   -{
182   - // Make a list of inheritable keys. Only the keys /MediaBox, /CropBox, /Resources, and /Rotate
183   - // are inheritable attributes. Push this object onto the stack of pages nodes that have values
184   - // for this attribute.
185   -
186   - std::set<std::string> inheritable_keys;
187   - for (auto const& key: cur_pages.getKeys()) {
188   - if (key == "/MediaBox" || key == "/CropBox" || key == "/Resources" || key == "/Rotate") {
189   - if (!allow_changes) {
190   - throw QPDFExc(
191   - qpdf_e_internal,
192   - m->file->getName(),
193   - m->last_object_description,
194   - m->file->getLastOffset(),
195   - "optimize detected an inheritable attribute when called in no-change mode");
196   - }
197   -
198   - // This is an inheritable resource
199   - inheritable_keys.insert(key);
200   - QPDFObjectHandle oh = cur_pages.getKey(key);
201   - QTC::TC("qpdf", "QPDF opt direct pages resource", oh.indirect() ? 0 : 1);
202   - if (!oh.indirect()) {
203   - if (!oh.isScalar()) {
204   - // Replace shared direct object non-scalar resources with indirect objects to
205   - // avoid copying large structures around.
206   - cur_pages.replaceKey(key, qpdf.makeIndirectObject(oh));
207   - oh = cur_pages.getKey(key);
208   - } else {
209   - // It's okay to copy scalars.
210   - }
211   - }
212   - key_ancestors[key].push_back(oh);
213   - if (key_ancestors[key].size() > 1) {
214   - }
215   - // Remove this resource from this node. It will be reattached at the page level.
216   - cur_pages.removeKey(key);
217   - } else if (!(key == "/Type" || key == "/Parent" || key == "/Kids" || key == "/Count")) {
218   - // Warn when flattening, but not if the key is at the top level (i.e. "/Parent" not
219   - // set), as we don't change these; but flattening removes intermediate /Pages nodes.
220   - if (warn_skipped_keys && cur_pages.hasKey("/Parent")) {
221   - warn(
222   - qpdf_e_pages,
223   - "Pages object: object " + cur_pages.id_gen().unparse(' '),
224   - 0,
225   - ("Unknown key " + key +
226   - " in /Pages object is being discarded as a result of flattening the /Pages "
227   - "tree"));
228   - }
229   - }
230   - }
231   -
232   - // Process descendant nodes. This method does not perform loop detection because all code paths
233   - // that lead here follow a call to getAllPages, which already throws an exception in the event
234   - // of a loop in the pages tree.
235   - for (auto& kid: cur_pages.getKey("/Kids").aitems()) {
236   - if (kid.isDictionaryOfType("/Pages")) {
237   - pushInheritedAttributesToPageInternal(
238   - kid, key_ancestors, allow_changes, warn_skipped_keys);
239   - } else {
240   - // Add all available inheritable attributes not present in this object to this object.
241   - for (auto const& iter: key_ancestors) {
242   - std::string const& key = iter.first;
243   - if (!kid.hasKey(key)) {
244   - kid.replaceKey(key, iter.second.back());
245   - } else {
246   - QTC::TC("qpdf", "QPDF opt page resource hides ancestor");
247   - }
248   - }
249   - }
250   - }
251   -
252   - // For each inheritable key, pop the stack. If the stack becomes empty, remove it from the map.
253   - // That way, the invariant that the list of keys in key_ancestors is exactly those keys for
254   - // which inheritable attributes are available.
255   -
256   - if (!inheritable_keys.empty()) {
257   - for (auto const& key: inheritable_keys) {
258   - key_ancestors[key].pop_back();
259   - if (key_ancestors[key].empty()) {
260   - key_ancestors.erase(key);
261   - }
262   - }
263   - } else {
264   - QTC::TC("qpdf", "QPDF opt no inheritable keys");
265   - }
266   -}
267   -
268   -void
269 138 Lin::updateObjectMaps(
270 139 ObjUser const& first_ou,
271 140 QPDFObjectHandle first_oh,
... ...
libqpdf/QPDF_pages.cc
  View file @ae90619
... ... @@ -280,6 +280,137 @@ Pages::flattenPagesTree()
280 280 }
281 281  
282 282 void
  283 +QPDF::pushInheritedAttributesToPage()
  284 +{
  285 + // Public API should not have access to allow_changes.
  286 + m->pages.pushInheritedAttributesToPage(true, false);
  287 +}
  288 +
  289 +void
  290 +Pages::pushInheritedAttributesToPage(bool allow_changes, bool warn_skipped_keys)
  291 +{
  292 + // Traverse pages tree pushing all inherited resources down to the page level.
  293 +
  294 + // The record of whether we've done this is cleared by updateAllPagesCache(). If we're warning
  295 + // for skipped keys, re-traverse unconditionally.
  296 + if (m->pushed_inherited_attributes_to_pages && (!warn_skipped_keys)) {
  297 + return;
  298 + }
  299 +
  300 + // Calling getAllPages() resolves any duplicated page objects, repairs broken nodes, and detects
  301 + // loops, so we don't have to do those activities here.
  302 + qpdf.getAllPages();
  303 +
  304 + // key_ancestors is a mapping of page attribute keys to a stack of Pages nodes that contain
  305 + // values for them.
  306 + std::map<std::string, std::vector<QPDFObjectHandle>> key_ancestors;
  307 + pushInheritedAttributesToPageInternal(
  308 + m->trailer.getKey("/Root").getKey("/Pages"),
  309 + key_ancestors,
  310 + allow_changes,
  311 + warn_skipped_keys);
  312 + if (!key_ancestors.empty()) {
  313 + throw std::logic_error(
  314 + "key_ancestors not empty after pushing inherited attributes to pages");
  315 + }
  316 + m->pushed_inherited_attributes_to_pages = true;
  317 + m->ever_pushed_inherited_attributes_to_pages = true;
  318 +}
  319 +
  320 +void
  321 +Pages::pushInheritedAttributesToPageInternal(
  322 + QPDFObjectHandle cur_pages,
  323 + std::map<std::string, std::vector<QPDFObjectHandle>>& key_ancestors,
  324 + bool allow_changes,
  325 + bool warn_skipped_keys)
  326 +{
  327 + // Make a list of inheritable keys. Only the keys /MediaBox, /CropBox, /Resources, and /Rotate
  328 + // are inheritable attributes. Push this object onto the stack of pages nodes that have values
  329 + // for this attribute.
  330 +
  331 + std::set<std::string> inheritable_keys;
  332 + for (auto const& key: cur_pages.getKeys()) {
  333 + if (key == "/MediaBox" || key == "/CropBox" || key == "/Resources" || key == "/Rotate") {
  334 + if (!allow_changes) {
  335 + throw QPDFExc(
  336 + qpdf_e_internal,
  337 + m->file->getName(),
  338 + m->last_object_description,
  339 + m->file->getLastOffset(),
  340 + "optimize detected an inheritable attribute when called in no-change mode");
  341 + }
  342 +
  343 + // This is an inheritable resource
  344 + inheritable_keys.insert(key);
  345 + QPDFObjectHandle oh = cur_pages.getKey(key);
  346 + QTC::TC("qpdf", "QPDF opt direct pages resource", oh.indirect() ? 0 : 1);
  347 + if (!oh.indirect()) {
  348 + if (!oh.isScalar()) {
  349 + // Replace shared direct object non-scalar resources with indirect objects to
  350 + // avoid copying large structures around.
  351 + cur_pages.replaceKey(key, qpdf.makeIndirectObject(oh));
  352 + oh = cur_pages.getKey(key);
  353 + } else {
  354 + // It's okay to copy scalars.
  355 + }
  356 + }
  357 + key_ancestors[key].push_back(oh);
  358 + if (key_ancestors[key].size() > 1) {
  359 + }
  360 + // Remove this resource from this node. It will be reattached at the page level.
  361 + cur_pages.removeKey(key);
  362 + } else if (!(key == "/Type" || key == "/Parent" || key == "/Kids" || key == "/Count")) {
  363 + // Warn when flattening, but not if the key is at the top level (i.e. "/Parent" not
  364 + // set), as we don't change these; but flattening removes intermediate /Pages nodes.
  365 + if (warn_skipped_keys && cur_pages.hasKey("/Parent")) {
  366 + warn(
  367 + qpdf_e_pages,
  368 + "Pages object: object " + cur_pages.id_gen().unparse(' '),
  369 + 0,
  370 + ("Unknown key " + key +
  371 + " in /Pages object is being discarded as a result of flattening the /Pages "
  372 + "tree"));
  373 + }
  374 + }
  375 + }
  376 +
  377 + // Process descendant nodes. This method does not perform loop detection because all code paths
  378 + // that lead here follow a call to getAllPages, which already throws an exception in the event
  379 + // of a loop in the pages tree.
  380 + for (auto& kid: cur_pages.getKey("/Kids").aitems()) {
  381 + if (kid.isDictionaryOfType("/Pages")) {
  382 + pushInheritedAttributesToPageInternal(
  383 + kid, key_ancestors, allow_changes, warn_skipped_keys);
  384 + } else {
  385 + // Add all available inheritable attributes not present in this object to this object.
  386 + for (auto const& iter: key_ancestors) {
  387 + std::string const& key = iter.first;
  388 + if (!kid.hasKey(key)) {
  389 + kid.replaceKey(key, iter.second.back());
  390 + } else {
  391 + QTC::TC("qpdf", "QPDF opt page resource hides ancestor");
  392 + }
  393 + }
  394 + }
  395 + }
  396 +
  397 + // For each inheritable key, pop the stack. If the stack becomes empty, remove it from the map.
  398 + // That way, the invariant that the list of keys in key_ancestors is exactly those keys for
  399 + // which inheritable attributes are available.
  400 +
  401 + if (!inheritable_keys.empty()) {
  402 + for (auto const& key: inheritable_keys) {
  403 + key_ancestors[key].pop_back();
  404 + if (key_ancestors[key].empty()) {
  405 + key_ancestors.erase(key);
  406 + }
  407 + }
  408 + } else {
  409 + QTC::TC("qpdf", "QPDF opt no inheritable keys");
  410 + }
  411 +}
  412 +
  413 +void
283 414 Pages::insertPageobjToPage(QPDFObjectHandle const& obj, int pos, bool check_duplicate)
284 415 {
285 416 QPDFObjGen og(obj.getObjGen());
... ...