0.7.3 start to look at document harvest (initial stub)

1 files changed, 186 insertions, 41 deletions

diff --git a/org/doc_reform.org b/org/doc_reform.org
index 12eb46a..7bb8029 100644
--- a/org/doc_reform.org
+++ b/org/doc_reform.org
@@ -27,7 +27,7 @@ struct Version {
   int minor;
   int patch;
 }
-enum _ver = Version(0, 7, 2);
+enum _ver = Version(0, 7, 3);
 #+END_SRC
 
 ** compilation restrictions (supported compilers)
@@ -71,8 +71,7 @@ version (Posix) {
 module doc_reform.sisu_document_parser;
 import
   doc_reform.conf.compile_time_info,
-  doc_reform.meta.metadoc,
-  doc_reform.meta.metadochead;
+  doc_reform.meta.metadoc;
 <<imports_doc_reform>>
 import std.algorithm;
 import std.parallelism;
@@ -86,6 +85,7 @@ homepage    "http://sisudoc.org"
 +/
 void main(string[] args) {
   <<doc_reform_mixin>>
+  <<doc_reform_init>>
   <<doc_reform_args>>
   <<doc_reform_do_selected>>
   if (_manifests.length > 1                            // _manifests[0] initialized dummy element
@@ -117,6 +117,38 @@ void main(string[] args) {
       }
     }
   }
+  if (_opt_action.verbose
+    && harvests.length > 1
+  ) {
+    auto min_repeat_number = 42;
+    foreach(doc_harvest; harvests) {
+      auto char_repeat_number = (doc_harvest.title.length
+        + doc_harvest.author.length + 16);
+      char_repeat_number = (char_repeat_number > min_repeat_number)
+      ? char_repeat_number
+      : min_repeat_number;
+      writefln(
+        "%s\n\"%s\", %s%s",
+        mkup.repeat_character_by_number_provided("-", char_repeat_number),
+        doc_harvest.title,
+        doc_harvest.author,
+        (doc_harvest.date_published.length > 0) ? " (" ~ doc_harvest.date_published ~ ")" : "",
+      );
+      string[] _topic_arr;
+      foreach(topic; doc_harvest.topic_register_arr.sort) {
+        foreach (i, _top; topic.split(mkup.sep)) {
+          writeln("  ", ("  ".repeat(i).join), "- ", _top);
+        }
+      }
+    }
+    string[] _author_date_title;
+    foreach(doc_harvest; harvests) {
+      _author_date_title ~= doc_harvest.author_date_title;
+    }
+    foreach(_adt; _author_date_title.sort) {
+      writeln(_adt);
+    }
+  }
 }
 #+END_SRC
 
@@ -193,6 +225,7 @@ import
 import
   doc_reform.meta,
   doc_reform.meta.metadoc_summary,
+  doc_reform.meta.metadoc_harvest,
   doc_reform.meta.metadoc_from_src,
   doc_reform.meta.conf_make_meta_structs,
   doc_reform.meta.conf_make_meta_toml,
@@ -266,7 +299,25 @@ mixin outputHub;
 
 **** init                                                           :init:
 
-#+NAME: doc_reform_args
+#+NAME: doc_reform_init
+#+BEGIN_SRC d
+struct Harvest {
+  string   title                = "";
+  string   author               = "";
+  string   author_date_title    = "";
+  string   date_published       = "";
+  string[] topic_register_arr   = [""];
+  string   html_seg_toc         = "";
+  string   html_scroll          = "";
+  string   epub                 = "";
+}
+Harvest harvested;
+Harvest[] harvests;
+#+END_SRC
+
+**** args                                                           :args:
+
+#+NAME: doc_reform_init
 #+BEGIN_SRC d
 string flag_action;
 string arg_unrecognized;
@@ -276,7 +327,7 @@ static auto rgx = Rgx();
 
 *** scope (run complete)                                            :scope:
 
-#+NAME: doc_reform_args
+#+NAME: doc_reform_init
 #+BEGIN_SRC d
 scope(success) {
   writefln(
@@ -311,6 +362,9 @@ bool[string] opts = [
   "debug"              : false,
   "digest"             : false,
   "epub"               : false,
+  "harvest"            : false,
+  "harvest-authors"    : false,
+  "harvest-topics"     : false,
   "html"               : false,
   "html-seg"           : false,
   "html-scroll"        : false,
@@ -366,6 +420,9 @@ auto helpInfo = getopt(args,
   "debug",              "--debug",                                                                  &opts["debug"],
   "digest",             "--digest hash digest for each object",                                     &opts["digest"],
   "epub",               "--epub process epub output",                                               &opts["epub"],
+  "harvest",            "--harvest extract info on authors & topics from document header metadata", &opts["harvest"],
+  "harvest-authors",    "--harvest-authors extract info on authors from document header metadata",  &opts["harvest-authors"],
+  "harvest-topics",     "--harvest-topics extract info on topics from document header metadata",    &opts["harvest-topics"],
   "html",               "--html process html output",                                               &opts["html"],
   "html-seg",           "--html-seg process html output",                                           &opts["html-seg"],
   "html-scroll",        "--html-seg process html output",                                           &opts["html-scroll"],
@@ -449,6 +506,22 @@ struct OptActions {
   bool epub() {
     return opts["epub"];
   }
+  bool harvest() {
+    bool _is = (
+      opts["harvest"]
+      || opts["harvest-authors"]
+      || opts["harvest-topics"]
+    )
+    ? true
+    : false;
+    return _is;
+  }
+  bool harvest_authors() {
+    return opts["harvest-authors"];
+  }
+  bool harvest_topics() {
+    return opts["harvest-topics"];
+  }
   bool html() {
     bool _is;
     if ( opts["html"] || opts["html-seg"] || opts["html-scroll"])
@@ -886,7 +959,29 @@ if ((doc_matters.opt.action.debug_do)
 #+BEGIN_SRC d
 /+ ↓ debugs +/
 if (doc_matters.opt.action.verbose) {
-  DocReformAbstractionSummary!()(doc_abstraction, doc_matters);
+  DocReformMetaDocSummary!()(doc_abstraction, doc_matters);
+}
+#+END_SRC
+
+**** 0. abstraction harvest                    :abstraction:harvest:
+- abstraction harvest
+
+#+NAME: doc_reform_each_file_do_debugs_checkdoc
+#+BEGIN_SRC d
+if (doc_matters.opt.action.harvest) {
+  if (doc_matters.opt.action.harvest_authors) {
+  }
+  if (doc_matters.opt.action.harvest_topics) {
+  }
+  Harvest[] DocReformMetaDocHarvests()(
+    Harvest    harvested,
+    Harvest[]  harvests,
+  ) {
+    harvests ~= harvested;
+    return harvests;
+  }
+  harvested = DocReformMetaDocHarvest!()(doc_matters, harvested);
+  harvests = DocReformMetaDocHarvests!()(harvested, harvests);
 }
 #+END_SRC
 
@@ -947,34 +1042,6 @@ break; // terminate, stop
 #+END_SRC
 
 * 2. pre-processing
-** Harvest _get document head_ for harvest (separate thread)
-*** 0 module template
-- harvest, get document head
-
-#+BEGIN_SRC d  :tangle "../src/doc_reform/meta/metadochead.d"
-module doc_reform.meta.metadochead;
-template DocReformHarvestGetFromHead() { // TODO
-  <<imports_doc_reform>>
-  <<doc_reform_mixin>>
-  enum headBody { header, body_content, insert_file_list, image_list }
-  enum makeMeta { make, meta }
-  static auto rgx = Rgx();
-  auto DocReformHarvestGetFromHead(E,O,M)( // TODO
-    E _env,
-    O _opt_action,
-    M _manifest
-  ){
-    <<doc_reform_conf_files_in_toml>>
-    <<doc_reform_each_file_do_read_and_split_sisu_markup_file_content_into_header_and_body>>
-    <<doc_reform_each_file_do_split_sisu_markup_file_header_into_make_and_meta_structs>>
-    <<doc_reform_each_file_do_document_matters_abridged>>
-    auto t = tuple(doc_matters_shared, doc_matters_abridged_collected);
-    static assert(t.length==2);
-    return t;
-  }
-}
-#+END_SRC
-
 ** Output _document abstraction functions_            :module:doc_reform:abstraction:
 *** 0 module template
 - abstraction template
@@ -1294,13 +1361,13 @@ if ((_opt_action.debug_do)
 #+END_SRC
 
 * 3. document abstraction _summary_         :module:doc_reform:metadoc_summary:
-** 0. module template
+** 0. module template metadoc summary
 - document summary from abstraction
 
 #+BEGIN_SRC d :tangle "../src/doc_reform/meta/metadoc_summary.d"
 module doc_reform.meta.metadoc_summary;
-template DocReformAbstractionSummary() {
-  void DocReformAbstractionSummary(S,T)(
+template DocReformMetaDocSummary() {
+  void DocReformMetaDocSummary(S,T)(
     const S  doc_abstraction,
           T  doc_matters,
   ) {
@@ -1308,7 +1375,7 @@ template DocReformAbstractionSummary() {
     mixin InternalMarkup;
     <<metadoc_summary_initialize>>
     if (doc_matters.opt.action.verbose) {
-      <<meta_metadoc_summary>>
+      <<meta_metadoc_summary_document>>
     }
   }
 }
@@ -1344,7 +1411,7 @@ auto markup = InlineMarkup();
 
 ** (last ocn)
 
-#+name: meta_metadoc_summary
+#+name: meta_metadoc_summary_document
 #+BEGIN_SRC d
 string[string] check = [
   "last_object_number" : "NA [debug \"checkdoc\" not run]",
@@ -1372,9 +1439,9 @@ foreach (k; doc_matters.has.keys_seq.seg) {
 }
 #+END_SRC
 
-** summary
+** document summary
 
-#+name: meta_metadoc_summary
+#+name: meta_metadoc_summary_document
 #+BEGIN_SRC d
 auto min_repeat_number = 66;
 auto char_repeat_number = (doc_matters.conf_make_meta.meta.title_full.length
@@ -1439,6 +1506,84 @@ writefln(
 );
 #+END_SRC
 
+** 0. module template metadoc harvest
+
+#+BEGIN_SRC d :tangle "../src/doc_reform/meta/metadoc_harvest.d"
+module doc_reform.meta.metadoc_harvest;
+template DocReformMetaDocHarvest() {
+  auto DocReformMetaDocHarvest(T,H)(
+    T  doc_matters,
+    H  harvest,
+  ) {
+    <<metadoc_harvest_imports>>
+    mixin InternalMarkup;
+    <<metadoc_harvest_initialize>>
+    <<meta_metadoc_harvest>>
+  }
+}
+#+END_SRC
+
+** init
+*** imports
+
+#+name: metadoc_harvest_imports
+#+BEGIN_SRC d
+import
+  doc_reform.meta.defaults,
+  doc_reform.meta.rgx;
+import
+  std.array,
+  std.exception,
+  std.regex,
+  std.stdio,
+  std.string,
+  std.traits,
+  std.typecons,
+  std.uni,
+  std.utf,
+  std.conv : to;
+#+END_SRC
+
+*** initialize                                                     :report:
+
+#+name: metadoc_harvest_initialize
+#+BEGIN_SRC d
+auto markup = InlineMarkup();
+#+END_SRC
+
+** harvest summary
+
+#+name: meta_metadoc_harvest_summary
+#+BEGIN_SRC d
+auto min_repeat_number = 66;
+auto char_repeat_number = (doc_matters.conf_make_meta.meta.title_full.length
+  + doc_matters.conf_make_meta.meta.creator_author.length + 4);
+char_repeat_number = (char_repeat_number > min_repeat_number)
+? char_repeat_number
+: min_repeat_number;
+writefln(
+  "%s\n\"%s\", %s\n%s\n%s\n%s",
+  markup.repeat_character_by_number_provided("-", char_repeat_number),
+  doc_matters.conf_make_meta.meta.title_full,
+  doc_matters.conf_make_meta.meta.creator_author,
+  doc_matters.src.filename,
+  doc_matters.conf_make_meta.meta.classify_topic_register_arr,
+  markup.repeat_character_by_number_provided("-", char_repeat_number),
+);
+#+END_SRC
+
+** return harvest
+
+#+name: meta_metadoc_harvest
+#+BEGIN_SRC d
+harvest.title              = doc_matters.conf_make_meta.meta.title_full;
+harvest.author             = doc_matters.conf_make_meta.meta.creator_author;
+harvest.author_date_title  = doc_matters.conf_make_meta.meta.author_date_title;
+harvest.date_published     = doc_matters.conf_make_meta.meta.date_published;
+harvest.topic_register_arr = doc_matters.conf_make_meta.meta.classify_topic_register_arr;
+return harvest;
+#+END_SRC
+
 * __END__
 dev notes