Metadata Harvest

From 0c4b3cb3707c3b16cd171620427e651d71182813 Mon Sep 17 00:00:00 2001 From: Ralph Amissah Date: Fri, 13 Dec 2019 09:09:24 -0500 Subject: present per document metadata --- src/doc_reform/io_out/hub.d | 6 + src/doc_reform/io_out/metadata.d | 385 +++++++++++++++++++++++++++++++++++ src/doc_reform/io_out/paths_output.d | 21 ++ src/doc_reform/io_out/xmls.d | 6 +- 4 files changed, 415 insertions(+), 3 deletions(-) create mode 100644 src/doc_reform/io_out/metadata.d (limited to 'src/doc_reform/io_out') diff --git a/src/doc_reform/io_out/hub.d b/src/doc_reform/io_out/hub.d index 8de07d8..a43c28a 100644 --- a/src/doc_reform/io_out/hub.d +++ b/src/doc_reform/io_out/hub.d @@ -5,6 +5,7 @@ module doc_reform.io_out.hub; template outputHub() { import doc_reform.io_out, + doc_reform.io_out.metadata, doc_reform.io_out.xmls, doc_reform.io_out.odt, doc_reform.io_out.create_zip_file, @@ -40,6 +41,10 @@ template outputHub() { doc_abstraction.outputEPub3!()(doc_matters); msg.vv("epub3 done"); } + if (sched == outTask.html_stuff) { + outputMetadata!()(doc_matters); + msg.vv("html metadata done"); + } if (sched == outTask.html_scroll) { msg.v("html scroll processing... "); import doc_reform.io_out.html; @@ -104,6 +109,7 @@ template outputHub() { } template outputHubOp() { import doc_reform.io_out, + doc_reform.io_out.metadata, doc_reform.io_out.xmls, doc_reform.io_out.odt, doc_reform.io_out.create_zip_file, diff --git a/src/doc_reform/io_out/metadata.d b/src/doc_reform/io_out/metadata.d new file mode 100644 index 0000000..f3b6176 --- /dev/null +++ b/src/doc_reform/io_out/metadata.d @@ -0,0 +1,385 @@ +module spine.io_out.metadata; +template outputMetadata() { + void outputMetadata(T)( T doc_matters) @safe { + import std.file; + import std.format; + import spine.io_out; + mixin InternalMarkup; + string[] metadata_; +string theme_dark_0 = format(q"┃ + body { + color : #CCCCCC; + background : #000000; + background-color : #000000; + } + a:link { + color : #FFFFFF; + text-decoration : none; + } + a:visited { + color : #999999; + text-decoration : none; + } + a:hover { + color : #000000; + background-color : #555555; + } + a:hover img { + background-color : #000000; + } + a:active { + color : #888888; + text-decoration : underline; + } +┃"); +string theme_light_0 = format(q"┃ + body { + color : #000000; + background : #FFFFFF; + background-color : #FFFFFF; + } + a:link { + color : #003399; + text-decoration : none; + } + a:visited { + color : #003399; + text-decoration : none; + } + a:hover { + color : #000000; + background-color : #f9f9aa; + } + a:hover img { + background-color : #FFFFFF; + } + a:active { + color : #003399; + text-decoration : underline; + } +┃"); +string theme_dark_1 = format(q"┃ + h1 { + color : #FFFFFF; + background : #000000; + } + p.letter { + color : #FFFFFF; + background : #333333; + } +┃"); +string theme_light_1 = format(q"┃ + h1 { + color : #FFFFFF; + background : #000088; + } + p.letter { + color : #FFFFFF; + background : #880000; + } +┃"); + metadata_ ~= format(q"┃ + + + +Metadata Harvest - Topics + + + + + + + + + +┃", + doc_matters.opt.action.css_theme_default ? theme_light_0 : theme_dark_0, + doc_matters.opt.action.css_theme_default ? theme_light_1 : theme_dark_1, +) ~ "\n"; + void metadata_write_output(M)(M doc_matters, string[] metadata_) @trusted { + auto pth_html = spinePathsHTML!()(doc_matters.output_path, doc_matters.src.language); + try { + if (!exists(pth_html.base)) { + pth_html.base.mkdirRecurse; + } + auto f = File(pth_html.fn_scroll("metadata." ~ doc_matters.src.filename), "w"); + foreach (o; metadata_) { + f.writeln(o); + } + } catch (ErrnoException ex) { + // Handle error + } + if (!(doc_matters.opt.action.quiet)) { + writeln(" ", pth_html.fn_scroll("metadata." ~ doc_matters.src.filename)); + } + } + static auto mkup = InlineMarkup(); + if (doc_matters.opt.action.debug_do) { + writeln(doc_matters.src.filename_base); + writeln("Title: ", doc_matters.conf_make_meta.meta.title_full); + writeln(" Author: ", doc_matters.conf_make_meta.meta.creator_author); + writeln(" Published: ", doc_matters.conf_make_meta.meta.date_published); + writeln(" Copyright: ", doc_matters.conf_make_meta.meta.rights_copyright); + writeln(" License: ", doc_matters.conf_make_meta.meta.rights_license); + if (doc_matters.conf_make_meta.meta.classify_topic_register_arr.length > 0) { + foreach (topic; doc_matters.conf_make_meta.meta.classify_topic_register_arr.sort!("toUpper(a) < toUpper(b)", SwapStrategy.unstable)) { + string[] subject_tree = topic.split(mkup.sep); + if (subject_tree.length > 0) { writeln(" ", subject_tree[0]); } + if (subject_tree.length > 1) { writeln(" ", subject_tree[1]); } + if (subject_tree.length > 2) { writeln(" ", subject_tree[2]); } + if (subject_tree.length > 3) { writeln(" ", subject_tree[3]); } + if (subject_tree.length > 4) { writeln(" ", subject_tree[4]); } + } + } + } + auto pth_html = spinePathsHTML!()(doc_matters.output_path, doc_matters.src.language); + auto pth_epub = spinePathsEPUB!()(doc_matters.output_path, doc_matters.src.language); + auto pth_pod = spinePathsPods!()(doc_matters); + metadata_ ~= format(q"┃ + + + + ┃"); + if (doc_matters.opt.action.harvest_link) { + metadata_ ~= format(q"┃

[ HOME ] Metadata Harvest + [ Authors ] + [ Topics ]

+ ┃"); + } else { + metadata_ ~= format(q"┃

[ HOME ] +

+ ┃"); + } + if (!(doc_matters.conf_make_meta.meta.title_full.empty)) { + metadata_ ~= "

Title: " ~ doc_matters.conf_make_meta.meta.title_full ~ "

"; + } else if (doc_matters.opt.action.debug_do || doc_matters.opt.action.very_verbose) { + writeln("ERROR no Title information provided in document header ", doc_matters.src.filename_base); + } + if (!(doc_matters.conf_make_meta.meta.creator_author.empty)) { + if (doc_matters.opt.action.harvest_link) { + metadata_ ~= "

Author: " + ~ doc_matters.conf_make_meta.meta.creator_author ~ "

"; + } else { + metadata_ ~= "

Author: " + ~ doc_matters.conf_make_meta.meta.creator_author ~ "

"; + } + } else if (doc_matters.opt.action.debug_do || doc_matters.opt.action.very_verbose) { + writeln("ERROR no Author information provided in document header ", doc_matters.src.filename_base); + } + metadata_ ~= "

Published: " ~ doc_matters.conf_make_meta.meta.date_published ~ "

"; + if (!(doc_matters.conf_make_meta.meta.rights_copyright.empty)) { + metadata_ ~= "

"; + } else if (doc_matters.opt.action.debug_do || doc_matters.opt.action.very_verbose) { + writeln("WARNING no Copyright information provided in document header ", doc_matters.src.filename_base); + } + if (!(doc_matters.conf_make_meta.meta.rights_license.empty)) { + metadata_ ~= "

License: " ~ doc_matters.conf_make_meta.meta.rights_license ~ "

"; + } else if (doc_matters.opt.action.debug_do || doc_matters.opt.action.very_verbose) { + writeln("WARNING no License information provided in document header ", doc_matters.src.filename_base); + } + if (!(doc_matters.conf_make_meta.meta.notes_summary.empty)) { + metadata_ ~= "

Summary:

" ~ doc_matters.conf_make_meta.meta.notes_summary ~ "

"; + } else if (doc_matters.opt.action.debug_do) { + writeln("WARNING no summary of text provided in document header ", doc_matters.src.filename_base); + } + if (doc_matters.conf_make_meta.meta.classify_topic_register_arr.length > 0) { + metadata_ ~= "

Topics:

"; + string[] _top = ["", "", "", "", ""]; + foreach (topic; doc_matters.conf_make_meta.meta.classify_topic_register_arr.sort!("toUpper(a) < toUpper(b)", SwapStrategy.unstable)) { + string[] subject_tree = topic.split(mkup.sep); + if (subject_tree.length > 0) { + if (subject_tree[0] != _top[0]) { + _top[0] = subject_tree[0]; + if (doc_matters.opt.action.harvest_link) { + metadata_ ~= + "

" + ~ subject_tree[0] + ~ "