digests (sha256) a revisitHEADmain

5 files changed, 87 insertions, 28 deletions

diff --git a/src/sisudoc/ocda/abstraction/ssp.d b/src/sisudoc/ocda/abstraction/ssp.d
index cb3747b..a2135df 100644
--- a/src/sisudoc/ocda/abstraction/ssp.d
+++ b/src/sisudoc/ocda/abstraction/ssp.d
@@ -291,14 +291,16 @@ template spineAbstractionTxt() {
         /+ ↓ sha256 digest +/
         {
           bool has_sha = false;
-          foreach (b; obj.metainfo.sha256) {
+          foreach (b; obj.metainfo.sha256.text) {
             if (b != 0) { has_sha = true; break; }
           }
           if (has_sha) {
-            output ~= ".sha256: " ~ obj.metainfo.sha256.toHexString.to!string;
+            if (doc_matters.opt.action.vox_gt_2) {
+              writeln(obj.metainfo.sha256.summary); // USE check
+            }
+            output ~= ".sha256: " ~ obj.metainfo.sha256.text.toHexString.to!string;
           }
         }
-
         /+ ↓ text attributes +/
         if (obj.attrib.indent_base != 0 || obj.attrib.indent_hang != 0)
           output ~= ".indent: " ~ obj.attrib.indent_base.to!string
diff --git a/src/sisudoc/ocda/meta/metadoc_from_src.d b/src/sisudoc/ocda/meta/metadoc_from_src.d
index 2955ee3..d41701c 100644
--- a/src/sisudoc/ocda/meta/metadoc_from_src.d
+++ b/src/sisudoc/ocda/meta/metadoc_from_src.d
@@ -1031,7 +1031,7 @@ template docAbstraction() {
            _images ~= extract_images(obj.text);
            obj = _image_dimensions(obj, manifested);
         }
-        obj.metainfo.sha256 = obj.obj_digest;
+        obj.metainfo.sha256 = obj.obj_digest(manifested);
         obj = _links(obj);
       }
     }
@@ -1103,7 +1103,7 @@ template docAbstraction() {
           obj.metainfo.ocn        = obj_cite_digits.object_number;
           obj.metainfo.identifier = obj_cite_digits.identifier;
         }
-        obj.metainfo.sha256 = obj.obj_digest;
+        obj.metainfo.sha256 = obj.obj_digest(manifested);
         obj = _links(obj);
       }
     }
@@ -1139,7 +1139,7 @@ template docAbstraction() {
           obj.metainfo.ocn                                = obj_cite_digits.object_number;
           obj.metainfo.identifier                         = obj_cite_digits.identifier;
         }
-        obj.metainfo.sha256 = obj.obj_digest;
+        obj.metainfo.sha256 = obj.obj_digest(manifested);
         obj = _links(obj);
       }
     }
@@ -1184,7 +1184,7 @@ template docAbstraction() {
           obj.metainfo.o_n_book_index                     = obj_cite_digits.bkidx;
           obj.metainfo.object_number_type                 = OCNtype.bkidx;
         }
-        obj.metainfo.sha256 = obj.obj_digest;
+        obj.metainfo.sha256 = obj.obj_digest(manifested);
         obj = _links(obj);
       }
       // TODO assert failure, reinstate
@@ -1222,7 +1222,7 @@ template docAbstraction() {
           obj.metainfo.object_number_off                  = obj_cite_digits.off;
           obj.metainfo.object_number_type                 = OCNtype.non;
         }
-        obj.metainfo.sha256 = obj.obj_digest;
+        obj.metainfo.sha256 = obj.obj_digest(manifested);
         obj = _links(obj);
       }
     }
@@ -1247,7 +1247,7 @@ template docAbstraction() {
       foreach (ref obj; the_document_head_section) {
         if (obj.metainfo.is_a == "heading") {
           if (obj.metainfo.heading_lev_markup == 0) { // sha for document root heading, caught here, should have been done earlier
-            obj.metainfo.sha256 = obj.obj_digest;
+            obj.metainfo.sha256 = obj.obj_digest(manifested);
           }
           foreach (pair; pairs) {
             if (obj.metainfo.ocn == pair[0]) {
diff --git a/src/sisudoc/ocda/meta/metadoc_from_src_functions.d b/src/sisudoc/ocda/meta/metadoc_from_src_functions.d
index f5c9bec..3bd9292 100644
--- a/src/sisudoc/ocda/meta/metadoc_from_src_functions.d
+++ b/src/sisudoc/ocda/meta/metadoc_from_src_functions.d
@@ -3184,10 +3184,50 @@ template docAbstractionFunctions() {
   }
   // ↑ - object tags
   // ↓ - object digest
-  pure ubyte[32] obj_digest()(
+  // pure
+  SHA_256_Hashes_ obj_digest(M)(
     ObjGenericComposite  obj,
+    M                    manifested,
   ) {
-    obj.metainfo.sha256 = obj.text.sha256Of;
+    ubyte[32] safeComputeHash(H)(H file_with_path) @trusted {
+      auto file_read = File(file_with_path, "r");
+      ubyte[32] file_sha256 = digest!SHA256(file_read.byChunk(4096 * 1024));
+      return file_sha256;
+    }
+    obj.metainfo.sha256.text = obj.text.sha256Of;
+    string image_with_path;
+    string image_path = manifested.src.image_dir_path ~ "/";
+    char[64] image_hash;
+    int n = 0;
+    if (obj.metainfo.is_a == "heading") {
+      obj.metainfo.sha256.summary ~=
+        obj.metainfo.sha256.text.toHexString ~ " "
+        ~ obj.metainfo.ocn.to!string ~ " "
+        ~ obj.metainfo.is_a ~ " "
+        ~ obj.metainfo.marked_up_level ~ "~ "
+        ~ obj.text;
+        // ~ image_summary; // heading assumed not to have image CONSIDER
+    } else {
+      if ((obj.text).match(rgx.inline_image)) {
+        foreach (m; (obj.text).matchAll(rgx.inline_image)) {
+          image_with_path = image_path ~ m["img"];
+          ubyte[32] image_sha256 = safeComputeHash(image_with_path);
+          obj.metainfo.sha256.images = [ m["img"] : image_sha256 ];
+        }
+      }
+      char[] image_summary;
+      if (obj.metainfo.sha256.images.length > 0) {
+        foreach (i,h; obj.metainfo.sha256.images) {
+          // writeln(h.toHexString.to!string, " - ", i);
+          image_summary ~= "\n" ~ h.toHexString.to!string ~ " - " ~  i;
+        }
+      }
+      obj.metainfo.sha256.summary ~=
+        obj.metainfo.sha256.text.toHexString
+        ~ " " ~ obj.metainfo.ocn.to!string ~ " "
+        ~ obj.metainfo.is_a
+        ~ image_summary;
+    }
     return obj.metainfo.sha256;
   }
   // ↑ - object digest
diff --git a/src/sisudoc/ocda/meta/metadoc_object_setter.d b/src/sisudoc/ocda/meta/metadoc_object_setter.d
index c5c8b1b..718ce55 100644
--- a/src/sisudoc/ocda/meta/metadoc_object_setter.d
+++ b/src/sisudoc/ocda/meta/metadoc_object_setter.d
@@ -101,6 +101,11 @@ template ObjectSetter() {
     string[]               lev4_subtoc                  = [];
     string[]               anchor_tags                  = [];
   }
+  struct SHA_256_Hashes_ {
+    char[]                 summary;
+    ubyte[32]              text;     // both text and inline footnotes; consider additionally separating text and footnotes
+    ubyte[32][string]      images;
+  }
   struct DocObj_MetaInfo_ {
     string      is_of_part          = ""; // frontmatter, body, backmatter
     string      is_of_section       = ""; // toc, body, glossary, biography, book index, blurb
@@ -119,7 +124,7 @@ template ObjectSetter() {
     int         o_n_bibliography    = 0;
     int         o_n_book_index      = 0;
     int         o_n_blurb           = 0;
-     string object_number_substantive() const @property {
+    string object_number_substantive() const @property {
       return (o_n_substantive == 0) ? "" : o_n_substantive.to!string;
     }
     string object_number_non_substantive() const @property {
@@ -174,7 +179,7 @@ template ObjectSetter() {
     int                    parent_ocn                           = 0;
     int                    last_descendant_ocn                  = 0;
     int[]                  children_headings;
-    ubyte[32]              sha256;
+    SHA_256_Hashes_        sha256;
   }
   struct ObjGenericComposite {
     string                 text         = "";
diff --git a/src/sisudoc/outputs/io_out/source_pod.d b/src/sisudoc/outputs/io_out/source_pod.d
index f6c73d6..b32abcd 100644
--- a/src/sisudoc/outputs/io_out/source_pod.d
+++ b/src/sisudoc/outputs/io_out/source_pod.d
@@ -51,6 +51,7 @@ module sisudoc.outputs.io_out.source_pod;
 @system: // is not @safe: use: @system: or @trusted:
 template spinePod() {
   import std.digest.sha;
+  // import std.digest : toHexString;
   import std.file;
   import std.outbuffer;
   import std.zip;
@@ -148,6 +149,21 @@ template spinePod() {
       ));
     }
   }
+  // ↓ - object digest & length
+  struct ST_file_stat {
+    ubyte[32] sha256hash;
+    ulong     fileSize;
+  }
+  ST_file_stat getFileStat(F)(F filepath) @trusted {
+    auto file = File(filepath, "r");
+    ubyte[32] file_sha256 = digest!SHA256(file.byChunk(4096 * 1024));
+    ulong fileSize = file.size;
+    ST_file_stat _file;
+    _file.sha256hash = file_sha256;
+    _file.fileSize = fileSize;
+    return _file;
+  }
+  // ↑ - object digest & length
   auto pod_zipMakeReady(M,P,S)(M doc_matters, P pths_pod, S _st) {
     auto pth_dr_doc_src = doc_matters.src_path_info;
     if (doc_matters.opt.action.vox_gt_3) { // correct
@@ -177,9 +193,8 @@ template spinePod() {
         if (exists(fn_src_in)) {
           debug(io) { writeln("(io debug) src out found: ", fn_src_in); }
           { // take DIGEST write to pod file digests.txt
-            auto data = (cast(byte[]) (fn_src_in).read);
-            _digests["shared"]["images"] ~= data.sha256Of.toHexString ~ "::" ~ data.length.to!string ~ " - " ~ image ~ "\n";
-            // writeln(data.sha256Of.toHexString, "::", data.length, " - ", image);
+            ST_file_stat file_ = getFileStat(fn_src_in);
+            _digests["shared"]["images"] ~= file_.sha256hash.toHexString ~ "::" ~ file_.fileSize.to!string ~ " - " ~ image ~ "\n";
           }
           if (doc_matters.opt.action.source_or_pod) {
             fn_src_in.copy(fn_src_out_filesystem);
@@ -216,9 +231,9 @@ template spinePod() {
             if (exists(fn_src_in)) {
               debug(io) { writeln("(io debug) src out found: ", fn_src_in); }
               { // take DIGEST write to pod file digests.txt
-                auto data = (cast(byte[]) (fn_src_in).read);
-                _digests[_lang]["ssp"] ~= data.sha256Of.toHexString
-                  ~ "::" ~ data.length.to!string ~ " - " ~ ssp_filename ~ "\n";
+                ST_file_stat file_ = getFileStat(fn_src_in);
+                _digests[_lang]["ssp"] ~= file_.sha256hash.toHexString
+                  ~ "::" ~ file_.fileSize.to!string ~ " - " ~ ssp_filename ~ "\n";
               }
               fn_src_in.copy(fn_src_out_filesystem);
               zip = podArchive("file_path_text", fn_src_in, fn_src_out_pod_zip_base, zip);
@@ -317,9 +332,8 @@ template spinePod() {
           if (exists(_pth_file_sstm)) { // what of language?
             debug(io) { writeln("(io debug) src in found: ", _pth_file_sstm); }
             { // take DIGEST write to pod file digests.txt
-              auto data = (cast(byte[]) (_pth_file_sstm).read);
-              _digests[_lang]["sstm"] ~= data.sha256Of.toHexString ~ "::" ~ data.length.to!string ~ " - " ~ doc_matters.src.filename ~ " - [" ~ _lang ~ "]";
-              // writeln(data.sha256Of.toHexString, "::", data.length, " - ", doc_matters.src.filename);
+              ST_file_stat file_ = getFileStat(_pth_file_sstm);
+              _digests[_lang]["sstm"] ~= file_.sha256hash.toHexString ~ "::" ~ file_.fileSize.to!string ~ " - " ~ doc_matters.src.filename ~ " - [" ~ _lang ~ "]";
             }
             filelist_src_out_pod_arr ~= fn_src_out_pod_zip_base;
             filelist_src_zpod_arr ~= fn_src_out_inside_pod;
@@ -378,9 +392,8 @@ template spinePod() {
               foreach (_lang; doc_matters.pod.manifest_list_of_languages) {
                 string _pth_mkup_src_in = _path_to_pod ~ _podname ~ "/" ~ _root_to_lang ~ _lang ~ "/" ~ _filename;
                 { // take DIGEST write to pod file digests.txt
-                  auto data = (cast(byte[]) (_pth_mkup_src_in).read);
-                  _digests[_language]["ssi"] ~= data.sha256Of.toHexString ~ "::" ~ data.length.to!string ~ " - " ~ _filename ~ " - [" ~ _lang ~ "]" ~ "\n";
-                  // writeln(data.sha256Of.toHexString, "::", data.length, " - ", _filename, " - [", _lang, "]");
+                  ST_file_stat file_ = getFileStat(_pth_mkup_src_in);
+                  _digests[_language]["ssi"] ~= file_.sha256hash.toHexString ~ "::" ~ file_.fileSize.to!string ~ " - " ~ _filename ~ " - [" ~ _lang ~ "]" ~ "\n";
                 }
                 string _pth_mkup_src_out = "pod/" ~ _root_to_lang ~ _lang ~ "/" ~ _filename;
                 if (exists(_pth_mkup_src_in)) {
@@ -406,9 +419,8 @@ template spinePod() {
           } else {
             auto fn_src_in = insert_file;
             { // take DIGEST write to pod file digests.txt // FIX likely bug insert_file includes path, not what you wish to write
-              auto data = (cast(byte[]) (fn_src_in).read);
-              _digests["en"]["ssi"] ~= data.sha256Of.toHexString ~ "::" ~ data.length.to!string ~ " - " ~ insert_file ~ "\n";
-              // writeln(data.sha256Of.toHexString, "::", data.length, " - ", insert_file);
+              ST_file_stat file_ = getFileStat(fn_src_in);
+              _digests["en"]["ssi"] ~= file_.sha256hash.toHexString ~ "::" ~ file_.fileSize.to!string ~ " - " ~ insert_file ~ "\n";
             }
             auto fn_src_out_pod_zip_base
               = pths_pod.fn_doc_insert(