From e95c49b76f4ac7bf72c383ee43a0567dfcbf1603 Mon Sep 17 00:00:00 2001 From: Ralph Amissah Date: Mon, 10 Sep 2018 18:15:02 -0400 Subject: 0.1.0 renamed doc-reform, doc_reform (& rad) - from sdp --- src/doc_reform/output/sqlite.d | 1242 ++++++++++++++++++++++++++++++++++++++++ 1 file changed, 1242 insertions(+) create mode 100644 src/doc_reform/output/sqlite.d (limited to 'src/doc_reform/output/sqlite.d') diff --git a/src/doc_reform/output/sqlite.d b/src/doc_reform/output/sqlite.d new file mode 100644 index 0000000..a765fd7 --- /dev/null +++ b/src/doc_reform/output/sqlite.d @@ -0,0 +1,1242 @@ +module doc_reform.output.sqlite; +import doc_reform.output; +import + std.file, + std.conv : to; +import d2sqlite3; +import std.typecons : Nullable; +mixin SiSUoutputRgxInit; +static auto rgx = Rgx(); +long _metadata_tid_lastrowid; +template SQLiteHubBuildTablesAndPopulate() { + void SQLiteHubBuildTablesAndPopulate(D,I)( + D doc_abstraction, + I doc_matters, + ) { + auto pth_sqlite = SiSUpathsSQLite!()(doc_matters.sqlite.filename, doc_matters.output_path); + pth_sqlite.base.mkdirRecurse; + auto db = Database(pth_sqlite.sqlite_file); + template SQLiteDbStatementComposite() { + void SQLiteDbStatementComposite(Db,D,I)( + Db db, + D doc_abstraction, + I doc_matters, + ) { + { + string _db_statement; + { + if ((doc_matters.opt.action.sqlite_db_create)) { + auto pth_sqlite = SiSUpathsSQLite!()(doc_matters.sqlite.filename, doc_matters.output_path); + pth_sqlite.base.mkdirRecurse; + _db_statement ~= SQLiteTablesReCreate!()(); + SQLiteDbRun!()(db, _db_statement, doc_matters.opt.action, "TABLE RE-CREATE"); + } + if (doc_matters.opt.action.sqlite_delete) { + _db_statement ~= SQLiteDeleteDocument!()(doc_matters); + SQLiteDbRun!()(db, _db_statement, doc_matters.opt.action, "DELETE Document"); + } + if (doc_matters.opt.action.sqlite_update) { + _db_statement ~= SQLiteDeleteDocument!()(doc_matters); + SQLiteDbRun!()(db, _db_statement, doc_matters.opt.action, "DELETE Document"); + _db_statement ~= SQLiteInsertMetadata!()(doc_matters); + SQLiteDbRun!()(db, _db_statement, doc_matters.opt.action, "INSERT MetaData"); + /+ get tid (lastrowid or max) for use in doc_objects table +/ + _db_statement ~= SQLiteInsertDocObjectsLoop!()(doc_abstraction, doc_matters); + SQLiteDbRun!()(db, _db_statement, doc_matters.opt.action, "INSERT DocObjects"); + } + } + db.close; + } + if (!(doc_matters.opt.action.quiet)) { + writeln(" ", pth_sqlite.sqlite_file); + } + } + } + SQLiteDbStatementComposite!()(db, doc_abstraction, doc_matters); + } +} +template SQLiteHubDiscreteBuildTablesAndPopulate() { + void SQLiteHubDiscreteBuildTablesAndPopulate(D,I)( + D doc_abstraction, + I doc_matters, + ) { + auto pth_sqlite = SiSUpathsSQLiteDiscrete!()(doc_matters.output_path, doc_matters.src.language); + pth_sqlite.base.mkdirRecurse; + auto db = Database(pth_sqlite.sqlite_file(doc_matters.src.filename)); + template SQLiteDiscreteDbStatementComposite() { + void SQLiteDiscreteDbStatementComposite(Db,D,I)( + Db db, + D doc_abstraction, + I doc_matters, + ) { + { + string _db_statement; + { + _db_statement ~= SQLiteTablesReCreate!()(); + _db_statement ~= SQLiteInsertMetadata!()(doc_matters); + _db_statement ~= SQLiteInsertDocObjectsLoop!()(doc_abstraction, doc_matters); + SQLiteDbRun!()(db, _db_statement, doc_matters.opt.action, "table CREATE Tables, INSERT DocObjects"); + } + db.close; + } + if (!(doc_matters.opt.action.quiet)) { + writeln(" ", pth_sqlite.sqlite_file(doc_matters.src.filename)); + } + } + } + SQLiteDiscreteDbStatementComposite!()(db, doc_abstraction, doc_matters); + } +} +template SQLiteDbRun() { + void SQLiteDbRun(Db,St,O)( + Db db, + St db_statement, + O opt_action, + string note, + ) { + debug(sql_statement) { + writeln(db_statement); + } + try { + db.run( + "\nBEGIN;\n" ~ + db_statement ~ + "\nCOMMIT;\n" + ); + if (!(opt_action.sqlite_discrete) + && !(opt_action.sqlite_db_create) + ) { + } + } catch (ErrnoException ex) { + writeln("ERROR SQLite : ", ex); + } catch (Exception ex) { + writeln("ERROR SQLite : ", ex); + } + if (opt_action.debug_do) { + writeln(note); + if (opt_action.verbose) { + writeln(db_statement); + } + } + } +} +template SQLinsertDelimiter() { + auto SQLinsertDelimiter(string _txt) { + _txt = _txt + .replaceAll(rgx.quotation_mark_sql_insert_delimiter, "$0$0"); + return _txt; + } +} +template SQLiteFormatAndLoadObject() { + auto SQLiteFormatAndLoadObject(I)( + I doc_matters, + ) { + mixin SiSUoutputRgxInit; + struct sqlite_format_and_load_objects { + auto generic_munge_sanitize_text_for_search( + string _txt, + ) { + string _notes; + string _urls; + if (_txt.matchFirst(rgx.inline_notes_al_gen)) { + foreach (m; _txt.matchAll(rgx.inline_notes_al_gen_text)) { + _notes ~= "\n" ~ m["text"]; + } + _txt = _txt.replaceAll(rgx.inline_notes_al_gen, ""); + } + if (_txt.matchFirst(rgx.inline_link)) { + foreach (m; _txt.matchAll(rgx.inline_link)) { + if (m["link"].match(rgx.url)) { + _urls ~= "\n" ~ m["link"]; + } + } + _txt = _txt.replaceAll(rgx.inline_link_clean, ""); + } + if (_notes.length > 0) { + _txt ~= _notes; + } + if (_urls.length > 0) { + _txt ~= _urls; + } + if (doc_matters.opt.action.debug_do) { + writeln(_txt, "\n"); + } + debug(sql_text_clean) { + writeln(_txt); + } + return _txt; + } + auto munge_html(O)( + auto ref const O obj, + ) { + string _html_special_characters(string _txt){ + _txt = _txt + .replaceAll(rgx.xhtml_ampersand, "&") + .replaceAll(rgx.xhtml_quotation, """) + .replaceAll(rgx.xhtml_less_than, "<") + .replaceAll(rgx.xhtml_greater_than, ">") + .replaceAll(rgx.nbsp_char, " ") + .replaceAll(rgx.xhtml_line_break, "
"); + return _txt; + } + string _html_font_face(string _txt){ + _txt = _txt + .replaceAll(rgx.inline_emphasis, "$1") + .replaceAll(rgx.inline_bold, "$1") + .replaceAll(rgx.inline_underscore, "$1") + .replaceAll(rgx.inline_italics, "$1") + .replaceAll(rgx.inline_superscript, "$1") + .replaceAll(rgx.inline_subscript, "$1") + .replaceAll(rgx.inline_strike, "$1") + .replaceAll(rgx.inline_insert, "$1") + .replaceAll(rgx.inline_mono, "$1") + .replaceAll(rgx.inline_cite, "$1"); + return _txt; + } + string _notes; + string _urls; + string _txt = _html_font_face(_html_special_characters(obj.text)); + if (_txt.matchFirst(rgx.inline_notes_al_gen)) { + foreach (m; _txt.matchAll(rgx.inline_notes_al_gen_text)) { + _notes ~= "\n" ~ m["text"]; + } + _txt = _txt.replaceAll(rgx.inline_notes_al_gen_ref, "$1 "); + } + if (_txt.matchFirst(rgx.inline_link)) { + foreach (m; _txt.matchAll(rgx.inline_link)) { + } + _txt = _txt.replaceAll(rgx.inline_link_clean, ""); + } + if (_notes.length > 0) { + _txt ~= _notes; + } + if (doc_matters.opt.action.debug_do) { + writeln(_txt, "\n"); + } + return _txt; + } + string html_special_characters(string _txt){ + _txt = _txt + .replaceAll(rgx.xhtml_ampersand, "&") + .replaceAll(rgx.xhtml_quotation, """) + .replaceAll(rgx.xhtml_less_than, "<") + .replaceAll(rgx.xhtml_greater_than, ">") + .replaceAll(rgx.nbsp_char, " ") + .replaceAll(rgx.xhtml_line_break, "
"); + return _txt; + } + string html_special_characters_code(string _txt){ + _txt = _txt + .replaceAll(rgx.xhtml_ampersand, "&") + .replaceAll(rgx.xhtml_quotation, """) + .replaceAll(rgx.xhtml_less_than, "<") + .replaceAll(rgx.xhtml_greater_than, ">") + .replaceAll(rgx.nbsp_char, " "); + return _txt; + } + string html_font_face(string _txt){ + _txt = _txt + .replaceAll(rgx.inline_emphasis, "$1") + .replaceAll(rgx.inline_bold, "$1") + .replaceAll(rgx.inline_underscore, "$1") + .replaceAll(rgx.inline_italics, "$1") + .replaceAll(rgx.inline_superscript, "$1") + .replaceAll(rgx.inline_subscript, "$1") + .replaceAll(rgx.inline_strike, "$1") + .replaceAll(rgx.inline_insert, "$1") + .replaceAll(rgx.inline_mono, "$1") + .replaceAll(rgx.inline_cite, "$1"); + return _txt; + } + auto html_heading(O)( + auto ref const O obj, + ) { + string _txt = munge_html(obj); + string o = format(q"¶

+ %s +

¶", + obj.metainfo.is_a, + _txt, + ); + return o; + } + auto html_para(O)( + auto ref const O obj, + ) { + string _txt = munge_html(obj); + _txt = (obj.attrib.bullet) ? ("●  " ~ _txt) : _txt; + string o = format(q"¶

+ %s +

¶", + obj.metainfo.is_a, + obj.attrib.indent_hang, + obj.attrib.indent_base, + _txt + ); + return o; + } + auto html_quote(O)( + auto ref const O obj, + ) { + string _txt = munge_html(obj); + string o = format(q"¶

+ %s +

¶", + obj.metainfo.is_a, + _txt + ); + return o; + } + auto html_group(O)( + auto ref const O obj, + ) { + string _txt = munge_html(obj); + string o = format(q"¶

+ %s +

¶", + obj.metainfo.is_a, + _txt + ); + return o; + } + auto html_block(O)( + auto ref const O obj, + ) { + string _txt = munge_html(obj); + string o = format(q"¶ +

%s

¶", + obj.metainfo.is_a, + _txt.stripRight + ); + return o; + } + auto html_verse(O)( + auto ref const O obj, + ) { + string _txt = munge_html(obj); + string o = format(q"¶

%s

¶", + obj.metainfo.is_a, + _txt + ); + return o; + } + auto html_code(O)( + auto ref const O obj, + ) { + string _txt = html_special_characters_code(obj.text); + string o = format(q"¶

%s

¶", + obj.metainfo.is_a, + _txt + ); + return o; + } + auto html_table(O)( + auto ref const O obj, + ) { + auto _tablarize(O)( + auto ref const O obj, + string _txt, + ) { + string[] _table_rows = _txt.split(rgx.table_delimiter_row); + string[] _table_cols; + string _table; + string _tablenote; + foreach(row_idx, row; _table_rows) { + _table_cols = row.split(rgx.table_delimiter_col); + _table ~= ""; + foreach(col_idx, cell; _table_cols) { + if ((_table_cols.length == 1) + && (_table_rows.length <= row_idx+2)) { // check row_idx+2 (rather than == ++row_idx) + _tablenote ~= cell; + } else { + string _col_is = (row_idx == 0 && obj.table.heading) ? "th" : "td"; + string _align = ("style=\"text-align:" + ~ ((obj.table.column_aligns[col_idx] == "l") + ? "left\"" : "right\"")); + _table ~= "<" + ~ _col_is + ~ " width=\"" + ~ obj.table.column_widths[col_idx].to!string + ~ "%\" " + ~ _align + ~ ">"; + _table ~= cell; + _table ~= ""; + } + } + _table ~= ""; + } + auto t = tuple( + _table, + _tablenote, + ); + return t; + } + string _txt = munge_html(obj); + auto t = _tablarize(obj, _txt); + _txt = t[0]; + string _note = t[1]; + string o = format(q"¶

+ + %s +
+ %s +

¶", + obj.metainfo.is_a, + _txt, + _note + ); + return o; + } + auto sqlite_load_string(O,Dm)( + auto ref const O obj, + auto ref Dm doc_matters, + ) { + string o; + return o; + } + auto postgresql_load_string(O,Dm)( + auto ref const O obj, + auto ref Dm doc_matters, + ) { + string o; + return o; + } + string sqlite_statement(O)( + auto ref const O obj, + string _txt, + string _html, + ) { + void _sql_exe(O)( + string _sql, + ) { + writeln(_html); + writeln(_sql); + } + string _sql; + return _sql; + } + auto heading(O)( + auto ref const O obj, + ) { + string[string] obj_txt = [ + "text": generic_munge_sanitize_text_for_search(obj.text), + "html": html_heading(obj) + ]; + if (doc_matters.opt.action.debug_do) { + debug(sql_txt) { + writeln(obj_txt["text"]); + } + debug(sql_html) { + writeln(obj_txt["html"]); + } + } else { + // load sql + } + return obj_txt; + } + auto para(O)( + auto ref const O obj, + ) { + string[string] obj_txt = [ + "text": generic_munge_sanitize_text_for_search(obj.text), + "html": html_para(obj) + ]; + if (doc_matters.opt.action.debug_do) { + debug(sql_txt) { + writeln(obj_txt["text"]); + } + debug(sql_html) { + writeln(obj_txt["html"]); + } + } else { + // load sql + } + return obj_txt; + } + auto quote(O)( + auto ref const O obj, + ) { + string[string] obj_txt = [ + "text": generic_munge_sanitize_text_for_search(obj.text), + "html": html_quote(obj) + ]; + if (doc_matters.opt.action.debug_do) { + debug(sql_txt) { + writeln(obj_txt["text"]); + } + debug(sql_html) { + writeln(obj_txt["html"]); + } + } else { + // load sql + } + return obj_txt; + } + auto group(O)( + auto ref const O obj, + ) { + string[string] obj_txt = [ + "text": generic_munge_sanitize_text_for_search(obj.text), + "html": html_group(obj) + ]; + if (doc_matters.opt.action.debug_do) { + debug(sql_txt) { + writeln(obj_txt["text"]); + } + debug(sql_html) { + writeln(obj_txt["html"]); + } + } else { + // load sql + } + return obj_txt; + } + auto block(O)( + auto ref const O obj, + ) { + string[string] obj_txt = [ + "text": generic_munge_sanitize_text_for_search(obj.text), + "html": html_block(obj) + ]; + if (doc_matters.opt.action.debug_do) { + debug(sql_txt) { + writeln(obj_txt["text"]); + } + debug(sql_html) { + writeln(obj_txt["html"]); + } + } else { + // load sql + } + return obj_txt; + } + auto verse(O)( + auto ref const O obj, + ) { + string[string] obj_txt = [ + "text": generic_munge_sanitize_text_for_search(obj.text), + "html": html_verse(obj) + ]; + if (doc_matters.opt.action.debug_do) { + debug(sql_txt) { + writeln(obj_txt["text"]); + } + debug(sql_html) { + writeln(obj_txt["html"]); + } + } else { + // load sql + } + return obj_txt; + } + auto code(O)( + auto ref const O obj, + ) { + string[string] obj_txt = [ + "text": generic_munge_sanitize_text_for_search(obj.text), + "html": html_code(obj) + ]; + if (doc_matters.opt.action.debug_do) { + debug(sql_txt) { + writeln(obj_txt["text"]); + } + debug(sql_html) { + writeln(obj_txt["html"]); + } + } else { + // load sql + } + return obj_txt; + } + auto table(O)( + auto ref const O obj, + ) { + string[string] obj_txt = [ + "text": generic_munge_sanitize_text_for_search(obj.text), + "html": html_table(obj) + ]; + if (doc_matters.opt.action.debug_do) { + debug(sql_txt) { + writeln(obj_txt["text"]); + } + debug(sql_html) { + writeln(obj_txt["html"]); + } + } else { + // load sql + } + return obj_txt; + } + } + return sqlite_format_and_load_objects(); + } +} +template SQLiteTablesReCreate() { + string SQLiteTablesReCreate()() { + string _sql_instruct; + _sql_instruct = format(q"¶ + DROP INDEX IF EXISTS idx_ocn; + DROP INDEX IF EXISTS idx_uid; + DROP INDEX IF EXISTS idx_digest_clean; + DROP INDEX IF EXISTS idx_digest_all; + DROP INDEX IF EXISTS idx_clean; + DROP INDEX IF EXISTS idx_title; + DROP INDEX IF EXISTS idx_creator_author; + DROP INDEX IF EXISTS src_filename; + DROP INDEX IF EXISTS idx_language_document_char; + DROP INDEX IF EXISTS idx_classify_topic_register; + DROP TABLE IF EXISTS metadata_and_text; + DROP TABLE IF EXISTS doc_objects; + DROP TABLE IF EXISTS urls; + CREATE TABLE metadata_and_text ( + uid VARCHAR(256) UNIQUE, /* filename, language char, pod/txt (decide on delimiter [,;:/]) */ + src_composite_id_per_txt VARCHAR(256) NOT NULL, /* UNIQUE, /* z pod name if any + src filename + language code */ + src_composite_id_per_pod VARCHAR(256) NOT NULL, /* z pod name if any + src filename */ + title VARCHAR(800) NOT NULL, + title_main VARCHAR(400) NOT NULL, + title_sub VARCHAR(400) NULL, + title_short VARCHAR(400) NULL, + title_edition VARCHAR(10) NULL, + title_note VARCHAR(2500) NULL, + title_language VARCHAR(100) NULL, + title_language_char VARCHAR(6) NULL, + creator_author VARCHAR(600) NOT NULL, + creator_author_email VARCHAR(100) NULL, + creator_author_hon VARCHAR(100) NULL, + creator_author_nationality VARCHAR(100) NULL, + creator_editor VARCHAR(600) NULL, + creator_contributor VARCHAR(600) NULL, + creator_illustrator VARCHAR(600) NULL, + creator_photographer VARCHAR(600) NULL, + creator_translator VARCHAR(600) NULL, + creator_prepared_by VARCHAR(600) NULL, + creator_digitized_by VARCHAR(600) NULL, + creator_audio VARCHAR(600) NULL, + creator_video VARCHAR(600) NULL, + language_document VARCHAR(100) NULL, + language_document_char VARCHAR(6) NOT NULL, + language_original VARCHAR(100) NULL, + language_original_char VARCHAR(6) NULL, + date_added_to_site VARCHAR(10) NULL, + date_available VARCHAR(10) NULL, + date_created VARCHAR(10) NULL, + date_issued VARCHAR(10) NULL, + date_modified VARCHAR(10) NULL, + date_published VARCHAR(10) NULL, + date_valid VARCHAR(10) NULL, + date_translated VARCHAR(10) NULL, + date_original_publication VARCHAR(10) NULL, + date_generated VARCHAR(10) NULL, + publisher VARCHAR(600) NULL, + original_publisher VARCHAR(600) NULL, + original_language VARCHAR(100) NULL, + original_language_char VARCHAR(6) NULL, + original_source VARCHAR(600) NULL, + original_institution VARCHAR(600) NULL, + original_nationality VARCHAR(100) NULL, + original_title VARCHAR(800) NULL, + rights_copyright VARCHAR(2500) NULL, + rights_copyright_audio VARCHAR(2500) NULL, + rights_copyright_cover VARCHAR(2500) NULL, + rights_copyright_illustrations VARCHAR(2500) NULL, + rights_copyright_photographs VARCHAR(2500) NULL, + rights_copyright_text VARCHAR(2500) NULL, + rights_copyright_translation VARCHAR(2500) NULL, + rights_copyright_video VARCHAR(2500) NULL, + rights_license VARCHAR(2500) NULL, + identifier_oclc VARCHAR(30) NULL, + identifier_isbn VARCHAR(16) NULL, + classify_topic_register VARCHAR(2500) NULL, + classify_subject VARCHAR(600) NULL, + classify_loc VARCHAR(30) NULL, + classify_dewey VARCHAR(30) NULL, + classify_keywords VARCHAR(600) NULL, + notes_abstract TEXT NULL, + notes_description TEXT NULL, + notes_comment TEXT NULL, + notes_coverage VARCHAR(200) NULL, + notes_relation VARCHAR(200) NULL, + notes_history VARCHAR(600) NULL, + notes_type VARCHAR(600) NULL, + notes_format VARCHAR(600) NULL, + notes_prefix TEXT NULL, + notes_prefix_a TEXT NULL, + notes_prefix_b TEXT NULL, + notes_suffix TEXT NULL, + src_filename VARCHAR(256) NOT NULL, + src_fingerprint VARCHAR(256) NULL, + src_filesize VARCHAR(10) NULL, + src_wordcount VARCHAR(10) NULL, + pod_name VARCHAR(256) NULL, /* zipped pod, work to be done here */ + pod_fingerprint VARCHAR(256) NULL, /* zipped pod, work to be done here */ + pod_size VARCHAR(10) NULL, /* zipped pod, work to be done here */ + src_text TEXT NULL, + fulltext TEXT NULL, + links TEXT NULL + ); + CREATE TABLE doc_objects ( + lid BIGINT PRIMARY KEY, + uid_metadata_and_text VARCHAR(256) REFERENCES metadata_and_text(uid) ON DELETE CASCADE, + ocn SMALLINT, + ocnd VARCHAR(6), + ocns VARCHAR(6), + clean TEXT NULL, + body TEXT NULL, + seg VARCHAR(256) NULL, + lev_an VARCHAR(1), + lev SMALLINT NULL, + is_of_type VARCHAR(16), + is_a VARCHAR(16), + node VARCHAR(16) NULL, + parent VARCHAR(16) NULL, + last_decendant VARCHAR(16) NULL, /* headings only */ + digest_clean CHAR(256), + digest_all CHAR(256), + types CHAR(1) NULL + ); + CREATE INDEX idx_ocn ON doc_objects(ocn); + CREATE INDEX idx_digest_clean ON doc_objects(digest_clean); + CREATE INDEX idx_digest_all ON doc_objects(digest_all); + CREATE INDEX idx_clean ON doc_objects(clean); + CREATE INDEX idx_title ON metadata_and_text(title); + CREATE INDEX idx_author ON metadata_and_text(creator_author); + CREATE INDEX idx_uid ON metadata_and_text(uid); + CREATE INDEX idx_filename ON metadata_and_text(src_filename); + CREATE INDEX idx_language ON metadata_and_text(language_document_char); + CREATE INDEX idx_topics ON metadata_and_text(classify_topic_register); + ¶",); + return _sql_instruct; + } +} +template SQLiteDeleteDocument() { + string SQLiteDeleteDocument(I)( + I doc_matters, + ) { + string _uid = doc_matters.src.doc_uid; + string _delete_uid = format(q"¶ + DELETE FROM metadata_and_text + WHERE uid = '%s'; + DELETE FROM doc_objects + WHERE uid_metadata_and_text = '%s'; + ¶", + _uid, + _uid, + ); + return _delete_uid; + } +} +template SQLiteInsertMetadata() { + string SQLiteInsertMetadata(I)( + I doc_matters, + ) { + string _uid = SQLinsertDelimiter!()(doc_matters.src.doc_uid); + string _insert_metadata = format(q"¶ + INSERT INTO metadata_and_text ( + uid, + src_filename, + src_composite_id_per_txt, + src_composite_id_per_pod, + title, + title_main, + title_sub, + title_short, + title_edition, + title_language, + classify_dewey, + classify_keywords, + classify_loc, + classify_subject, + classify_topic_register, + creator_author, + creator_author_email, + creator_illustrator, + creator_translator, + date_added_to_site, + date_available, + date_created, + date_issued, + date_modified, + date_published, + date_valid, + identifier_isbn, + identifier_oclc, + language_document, + language_document_char, + notes_abstract, + notes_description, + original_publisher, + original_language, + original_language_char, + original_source, + original_title, + publisher, + rights_copyright, + rights_copyright_audio, + rights_copyright_cover, + rights_copyright_illustrations, + rights_copyright_photographs, + rights_copyright_text, + rights_copyright_translation, + rights_copyright_video, + rights_license + ) + VALUES ( + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s', + '%s' + ); + ¶", + _uid, + SQLinsertDelimiter!()(doc_matters.src.filename), + SQLinsertDelimiter!()(doc_matters.src.docname_composite_unique_per_src_doc), + SQLinsertDelimiter!()(doc_matters.src.docname_composite_unique_per_src_pod), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.title_full), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.title_main), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.title_subtitle), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.title_short), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.title_edition), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.title_language), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.classify_dewey), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.classify_keywords), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.classify_loc), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.classify_subject), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.classify_topic_register), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.creator_author), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.creator_author_email), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.creator_illustrator), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.creator_translator), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.date_added_to_site), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.date_available), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.date_created), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.date_issued), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.date_modified), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.date_published), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.date_valid), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.identifier_isbn), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.identifier_oclc), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.language_document), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.language_document_char), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.notes_abstract), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.notes_description), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.original_publisher), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.original_language), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.original_language_char), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.original_source), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.original_title), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.publisher), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.rights_copyright), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.rights_copyright_audio), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.rights_copyright_cover), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.rights_copyright_illustrations), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.rights_copyright_photographs), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.rights_copyright_text), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.rights_copyright_translation), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.rights_copyright_video), + SQLinsertDelimiter!()(doc_matters.conf_make_meta.meta.rights_license) + ); + return _insert_metadata; + } +} +template SQLiteInsertDocObjectsLoop() { + string SQLiteInsertDocObjectsLoop(D,I)( + D doc_abstraction, + I doc_matters, + ) { + string _uid = SQLinsertDelimiter!()(doc_matters.src.doc_uid); + string insertDocObjectsRow(O)(O obj) { + string _insert_doc_objects_row = format(q"¶ + INSERT INTO doc_objects ( + uid_metadata_and_text, + ocn, + clean, + body, + lev, + is_of_type, + is_a + ) + VALUES ( + '%s', + %s, + '%s', + '%s', + %s, + '%s', + '%s' + ); + ¶", + _uid, + obj.metainfo.ocn, + SQLinsertDelimiter!()(obj_txt["text"]), + SQLinsertDelimiter!()(obj_txt["html"]), + obj.metainfo.heading_lev_markup, + obj.metainfo.is_of_type, + obj.metainfo.is_a, + ); + return _insert_doc_objects_row; + } + auto format_and_sqlite_load = SQLiteFormatAndLoadObject!()(doc_matters); + string[string] obj_txt; + string doc_text; + string[] _insert_doc_objects; + foreach (part; doc_matters.xml.keys_seq.sql) { + foreach (obj; doc_abstraction[part]) { + switch (obj.metainfo.is_of_part) { + case "frontmatter": assert(part == "head", part); + switch (obj.metainfo.is_of_type) { + case "para": + switch (obj.metainfo.is_a) { + case "heading": + obj_txt = format_and_sqlite_load.heading(obj); + break; + default: + if ((doc_matters.opt.action.debug_do)) { + writeln(__FILE__, ":", __LINE__, ": ", obj.metainfo.is_a); + } + break; + } + break; + default: + if ((doc_matters.opt.action.debug_do)) { + writeln(__FILE__, ":", __LINE__, ": ", obj.metainfo.is_of_type); + } + break; + } + break; + case "body": // assert(part == "body", part); + switch (obj.metainfo.is_of_type) { + case "para": + switch (obj.metainfo.is_a) { + case "heading": + debug (asserts) { + if (part != "body") { + writeln(__LINE__, ": ", obj.text); + } + } + obj_txt = format_and_sqlite_load.heading(obj); + break; + case "para": + obj_txt = format_and_sqlite_load.para(obj); + break; + default: + if ((doc_matters.opt.action.debug_do)) { + writeln(__FILE__, ":", __LINE__, ": ", obj.metainfo.is_a); + } + break; + } + break; + case "block": + switch (obj.metainfo.is_a) { + case "quote": + obj_txt = format_and_sqlite_load.quote(obj); + break; + case "group": + obj_txt = format_and_sqlite_load.group(obj); + break; + case "block": + obj_txt = format_and_sqlite_load.block(obj); + break; + case "poem": // double check on keeping both poem & verse + break; + case "verse": + obj_txt = format_and_sqlite_load.verse(obj); + break; + case "code": + obj_txt = format_and_sqlite_load.code(obj); + break; + case "table": + obj_txt = format_and_sqlite_load.table(obj); + break; + default: + if ((doc_matters.opt.action.debug_do)) { + writeln(__FILE__, ":", __LINE__, ": ", obj.metainfo.is_a); + } + break; + } + break; + default: + if ((doc_matters.opt.action.debug_do)) { + writeln(__FILE__, ":", __LINE__, ": ", obj.metainfo.is_of_type); + } + break; + } + break; + case "backmatter": + assert(part == "glossary" || "bibliography" || "bookindex_seg" || "blurb" || "tail", part); + switch (obj.metainfo.is_of_type) { + case "para": + switch (obj.metainfo.is_a) { + case "heading": + obj_txt = format_and_sqlite_load.heading(obj); + break; + case "glossary": assert(part == "glossary", part); + obj_txt = format_and_sqlite_load.para(obj); + break; + case "bibliography": assert(part == "bibliography", part); + obj_txt = format_and_sqlite_load.para(obj); + break; + case "bookindex": assert(part == "bookindex_seg", part); + obj_txt = format_and_sqlite_load.para(obj); + break; + case "blurb": assert(part == "blurb", part); + obj_txt = format_and_sqlite_load.para(obj); + break; + default: + if ((doc_matters.opt.action.debug_do)) { + writeln(__FILE__, ":", __LINE__, ": ", obj.metainfo.is_a); + } + break; + } + break; + default: + if ((doc_matters.opt.action.debug_do)) { + writeln(__FILE__, ":", __LINE__, ": ", obj.metainfo.is_of_type); + } + break; + } + break; + case "comment": + break; + default: + if ((doc_matters.opt.action.debug_do)) { + writeln(__FILE__, ":", __LINE__, ": ", obj.metainfo.is_of_part); // check where empty value could come from + writeln(__FILE__, ":", __LINE__, ": ", obj.metainfo.is_a); + writeln(__FILE__, ":", __LINE__, ": ", obj.text); // check where empty value could come from + } + break; + } + if (obj.metainfo.is_a == "heading") { + if ((doc_matters.opt.action.very_verbose)) { + writeln( + "markup: ", obj.metainfo.heading_lev_markup, + "> ", obj.metainfo.dom_structure_markedup_tags_status, + "; collapsed: ", obj.metainfo.heading_lev_collapsed, + "> ", obj.metainfo.dom_structure_collapsed_tags_status, + "; ocn: ", obj.metainfo.ocn, + " node: ", obj.metainfo.node, + "; parent: ", obj.metainfo.parent_lev_markup, + "; ocn: ", obj.metainfo.parent_ocn, + "; ", + ); + } + } + if (!(obj.metainfo.is_a == "comment")) { + _insert_doc_objects ~= insertDocObjectsRow(obj); + } + } // loop closes + } + return _insert_doc_objects.join.to!(char[]).toUTF8; + } +} +template SQLiteTablesCreate() { + void SQLiteTablesCreate(E,O)(E env, O opt_action) { + import d2sqlite3; + template SQLiteTablesReCreate() { + string SQLiteTablesReCreate()() { + string _sql_instruct; + _sql_instruct = format(q"¶ + DROP INDEX IF EXISTS idx_ocn; + DROP INDEX IF EXISTS idx_uid; + DROP INDEX IF EXISTS idx_digest_clean; + DROP INDEX IF EXISTS idx_digest_all; + DROP INDEX IF EXISTS idx_clean; + DROP INDEX IF EXISTS idx_title; + DROP INDEX IF EXISTS idx_creator_author; + DROP INDEX IF EXISTS src_filename; + DROP INDEX IF EXISTS idx_language_document_char; + DROP INDEX IF EXISTS idx_classify_topic_register; + DROP TABLE IF EXISTS metadata_and_text; + DROP TABLE IF EXISTS doc_objects; + DROP TABLE IF EXISTS urls; + CREATE TABLE metadata_and_text ( + uid VARCHAR(256) UNIQUE, /* filename, language char, pod/txt (decide on delimiter [,;:/]) */ + src_composite_id_per_txt VARCHAR(256) NOT NULL, /* UNIQUE, /* z pod name if any + src filename + language code */ + src_composite_id_per_pod VARCHAR(256) NOT NULL, /* z pod name if any + src filename */ + title VARCHAR(800) NOT NULL, + title_main VARCHAR(400) NOT NULL, + title_sub VARCHAR(400) NULL, + title_short VARCHAR(400) NULL, + title_edition VARCHAR(10) NULL, + title_note VARCHAR(2500) NULL, + title_language VARCHAR(100) NULL, + title_language_char VARCHAR(6) NULL, + creator_author VARCHAR(600) NOT NULL, + creator_author_email VARCHAR(100) NULL, + creator_author_hon VARCHAR(100) NULL, + creator_author_nationality VARCHAR(100) NULL, + creator_editor VARCHAR(600) NULL, + creator_contributor VARCHAR(600) NULL, + creator_illustrator VARCHAR(600) NULL, + creator_photographer VARCHAR(600) NULL, + creator_translator VARCHAR(600) NULL, + creator_prepared_by VARCHAR(600) NULL, + creator_digitized_by VARCHAR(600) NULL, + creator_audio VARCHAR(600) NULL, + creator_video VARCHAR(600) NULL, + language_document VARCHAR(100) NULL, + language_document_char VARCHAR(6) NOT NULL, + language_original VARCHAR(100) NULL, + language_original_char VARCHAR(6) NULL, + date_added_to_site VARCHAR(10) NULL, + date_available VARCHAR(10) NULL, + date_created VARCHAR(10) NULL, + date_issued VARCHAR(10) NULL, + date_modified VARCHAR(10) NULL, + date_published VARCHAR(10) NULL, + date_valid VARCHAR(10) NULL, + date_translated VARCHAR(10) NULL, + date_original_publication VARCHAR(10) NULL, + date_generated VARCHAR(10) NULL, + publisher VARCHAR(600) NULL, + original_publisher VARCHAR(600) NULL, + original_language VARCHAR(100) NULL, + original_language_char VARCHAR(6) NULL, + original_source VARCHAR(600) NULL, + original_institution VARCHAR(600) NULL, + original_nationality VARCHAR(100) NULL, + original_title VARCHAR(800) NULL, + rights_copyright VARCHAR(2500) NULL, + rights_copyright_audio VARCHAR(2500) NULL, + rights_copyright_cover VARCHAR(2500) NULL, + rights_copyright_illustrations VARCHAR(2500) NULL, + rights_copyright_photographs VARCHAR(2500) NULL, + rights_copyright_text VARCHAR(2500) NULL, + rights_copyright_translation VARCHAR(2500) NULL, + rights_copyright_video VARCHAR(2500) NULL, + rights_license VARCHAR(2500) NULL, + identifier_oclc VARCHAR(30) NULL, + identifier_isbn VARCHAR(16) NULL, + classify_topic_register VARCHAR(2500) NULL, + classify_subject VARCHAR(600) NULL, + classify_loc VARCHAR(30) NULL, + classify_dewey VARCHAR(30) NULL, + classify_keywords VARCHAR(600) NULL, + notes_abstract TEXT NULL, + notes_description TEXT NULL, + notes_comment TEXT NULL, + notes_coverage VARCHAR(200) NULL, + notes_relation VARCHAR(200) NULL, + notes_history VARCHAR(600) NULL, + notes_type VARCHAR(600) NULL, + notes_format VARCHAR(600) NULL, + notes_prefix TEXT NULL, + notes_prefix_a TEXT NULL, + notes_prefix_b TEXT NULL, + notes_suffix TEXT NULL, + src_filename VARCHAR(256) NOT NULL, + src_fingerprint VARCHAR(256) NULL, + src_filesize VARCHAR(10) NULL, + src_wordcount VARCHAR(10) NULL, + pod_name VARCHAR(256) NULL, /* zipped pod, work to be done here */ + pod_fingerprint VARCHAR(256) NULL, /* zipped pod, work to be done here */ + pod_size VARCHAR(10) NULL, /* zipped pod, work to be done here */ + src_text TEXT NULL, + fulltext TEXT NULL, + links TEXT NULL + ); + CREATE TABLE doc_objects ( + lid BIGINT PRIMARY KEY, + uid_metadata_and_text VARCHAR(256) REFERENCES metadata_and_text(uid) ON DELETE CASCADE, + ocn SMALLINT, + ocnd VARCHAR(6), + ocns VARCHAR(6), + clean TEXT NULL, + body TEXT NULL, + seg VARCHAR(256) NULL, + lev_an VARCHAR(1), + lev SMALLINT NULL, + is_of_type VARCHAR(16), + is_a VARCHAR(16), + node VARCHAR(16) NULL, + parent VARCHAR(16) NULL, + last_decendant VARCHAR(16) NULL, /* headings only */ + digest_clean CHAR(256), + digest_all CHAR(256), + types CHAR(1) NULL + ); + CREATE INDEX idx_ocn ON doc_objects(ocn); + CREATE INDEX idx_digest_clean ON doc_objects(digest_clean); + CREATE INDEX idx_digest_all ON doc_objects(digest_all); + CREATE INDEX idx_clean ON doc_objects(clean); + CREATE INDEX idx_title ON metadata_and_text(title); + CREATE INDEX idx_author ON metadata_and_text(creator_author); + CREATE INDEX idx_uid ON metadata_and_text(uid); + CREATE INDEX idx_filename ON metadata_and_text(src_filename); + CREATE INDEX idx_language ON metadata_and_text(language_document_char); + CREATE INDEX idx_topics ON metadata_and_text(classify_topic_register); + ¶",); + return _sql_instruct; + } + } + if (opt_action.sqlite_db_create) { + string _db_statement; + auto pth_sqlite = SiSUpathsSQLite!()(opt_action.sqlite_filename, opt_action.output_dir_set); + pth_sqlite.base.mkdirRecurse; + auto db = Database(pth_sqlite.sqlite_file); + { + _db_statement ~= SQLiteTablesReCreate!()(); + } + SQLiteDbRun!()(db, _db_statement, opt_action, "TABLE RE-CREATE"); + } + } +} +template SQLiteDbDrop() { + void SQLiteDbDrop(O)(O opt_action) { + writeln("db drop"); + if ((opt_action.sqlite_db_drop)) { + auto pth_sqlite = SiSUpathsSQLite!()(opt_action.sqlite_filename, opt_action.output_dir_set); + writeln("remove(", pth_sqlite.sqlite_file, ")"); + try { + remove(pth_sqlite.sqlite_file); + } catch (FileException ex) { + // handle error + } + } + } +} -- cgit v1.2.3