0.7.3 start to look at document harvest (initial stub)

1 files changed, 100 insertions, 4 deletions

diff --git a/src/doc_reform/doc_reform.d b/src/doc_reform/doc_reform.d
index 06866c2..908a3a6 100755
--- a/src/doc_reform/doc_reform.d
+++ b/src/doc_reform/doc_reform.d
@@ -53,8 +53,7 @@
 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;
 import
   std.datetime,
   std.getopt,
@@ -64,6 +63,7 @@ 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,
@@ -91,6 +91,18 @@ void main(string[] args) {
   mixin DocReformBiblio;
   mixin DocReformRgxInitFlags;
   mixin outputHub;
+  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;
   string flag_action;
   string arg_unrecognized;
   enum dAM { abstraction, matters }
@@ -119,6 +131,9 @@ void main(string[] args) {
     "debug"              : false,
     "digest"             : false,
     "epub"               : false,
+    "harvest"            : false,
+    "harvest-authors"    : false,
+    "harvest-topics"     : false,
     "html"               : false,
     "html-seg"           : false,
     "html-scroll"        : false,
@@ -174,6 +189,9 @@ void main(string[] 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"],
@@ -250,6 +268,22 @@ void main(string[] args) {
     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"])
@@ -637,7 +671,22 @@ void main(string[] args) {
           }
           /+ ↓ debugs +/
           if (doc_matters.opt.action.verbose) {
-            DocReformAbstractionSummary!()(doc_abstraction, doc_matters);
+            DocReformMetaDocSummary!()(doc_abstraction, doc_matters);
+          }
+          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);
           }
           /+ ↓ debugs +/
           if (doc_matters.opt.action.debug_do) {
@@ -717,7 +766,22 @@ void main(string[] args) {
           }
           /+ ↓ debugs +/
           if (doc_matters.opt.action.verbose) {
-            DocReformAbstractionSummary!()(doc_abstraction, doc_matters);
+            DocReformMetaDocSummary!()(doc_abstraction, doc_matters);
+          }
+          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);
           }
           /+ ↓ debugs +/
           if (doc_matters.opt.action.debug_do) {
@@ -755,4 +819,36 @@ 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);
+    }
+  }
 }