diff options
Diffstat (limited to 'src/sisudoc/meta/rgx.d')
| -rw-r--r-- | src/sisudoc/meta/rgx.d | 32 |
1 files changed, 21 insertions, 11 deletions
diff --git a/src/sisudoc/meta/rgx.d b/src/sisudoc/meta/rgx.d index 0b5f9f0..86ca40c 100644 --- a/src/sisudoc/meta/rgx.d +++ b/src/sisudoc/meta/rgx.d @@ -6,7 +6,7 @@ - Author: Ralph Amissah [ralph.amissah@gmail.com] - - Copyright: (C) 2015 - 2024 Ralph Amissah, All Rights Reserved. + - Copyright: (C) 2015 - 2025 Ralph Amissah, All Rights Reserved. - License: AGPL 3 or later: @@ -108,8 +108,8 @@ static template spineRgxIn() { static para_attribs = ctRegex!(`^_(?:(?:[0-9])(?:_([0-9]))?|(?:[1-9])?[*]) `); static para_inline_link_anchor = ctRegex!(`\*[~](?P<anchor>[a-z0-9_.-]+)(?= |$)`,"i"); /+ blocked markup +/ - static block_open = ctRegex!("^((code(?:[.][a-z][0-9a-z#+_]+)?|(?:poem|group|block|quote)(?:[.][a-z][0-9a-z_]+)?|table)(?:[(][ a-zA-Z0-9;:,]*[)])?[{][ ]*$)|^`{3} (code(?:[.][a-z][0-9a-z#+_]+)?|(?:poem|group|block|quote)(?:[.][a-z][0-9a-z_]+)?|table)(?:[(][ a-zA-Z0-9;:,]*[)])?|^[{]table[(](?:h;)?(?P<columns>(?:[ ,]+[0-9]+)+)[)][}]"); - static block_poem_open = ctRegex!("^((poem(?:[(][ a-zA-Z0-9;:,]*[)])?[{][ ]*$)|`{3} poem(?:[(][ a-zA-Z0-9;:,]*[)])?)"); + static block_open = ctRegex!("^((code(?:[.][a-z][0-9a-z#+_]+)?|(?:poem|group|block|quote)(?:[.][a-z][0-9a-z_]+)?|table)(?:[(][ a-zA-Z0-9;:,]*[)])?[{][ ]*$)|^[`']{3} (code(?:[.][a-z][0-9a-z#+_]+)?|(?:poem|group|block|quote)(?:[.][a-z][0-9a-z_]+)?|table)(?:[(][ a-zA-Z0-9;:,]*[)])?|^[{]table[(](?:h;)?(?P<columns>(?:[ ,]+[0-9]+)+)[)][}]"); + static block_poem_open = ctRegex!("^((poem(?:[(][ a-zA-Z0-9;:,]*[)])?[{][ ]*$)|[`']{3} poem(?:[(][ a-zA-Z0-9;:,]*[)])?)"); /+ blocked markup tics +/ static block_tic_code_open = ctRegex!("^`{3} code(?:[.](?P<syntax>[a-z][0-9a-z#+_]+))?(?:[(](?P<attrib>[ a-zA-Z0-9;:,]*)[)])?"); static block_tic_poem_open = ctRegex!("^`{3} poem(?:[.](?P<lang>[a-z][0-9a-z_]+))?(?:[(](?P<attrib>[ a-zA-Z0-9;:,]*)[)])?"); @@ -118,6 +118,14 @@ static template spineRgxIn() { static block_tic_quote_open = ctRegex!("^`{3} quote(?:[.](?P<lang>[a-z][0-9a-z_]+))?(?:[(](?P<attrib>[ a-zA-Z0-9;:,]*)[)])?"); static block_tic_table_open = ctRegex!("^`{3} table(?:[(](?P<attrib>[ a-zA-Z0-9;:,]*)[)])?"); // ctRegex!("^`{3} table(?:\(.*?\))?"); static block_tic_close = ctRegex!("^(`{3})$","m"); + /+ blocked markup tics +/ + static block_quotemarks_code_open = ctRegex!(`^'{3} code(?:[.](?P<syntax>[a-z][0-9a-z#+_]+))?(?:[(](?P<attrib>[ a-zA-Z0-9;:,]*)[)])?`); + static block_quotemarks_poem_open = ctRegex!(`^'{3} poem(?:[.](?P<lang>[a-z][0-9a-z_]+))?(?:[(](?P<attrib>[ a-zA-Z0-9;:,]*)[)])?`); + static block_quotemarks_group_open = ctRegex!(`^'{3} group(?:[.](?P<lang>[a-z][0-9a-z_]+))?(?:[(](?P<attrib>[ a-zA-Z0-9;:,]*)[)])?`); + static block_quotemarks_block_open = ctRegex!(`^'{3} block(?:[.](?P<lang>[a-z][0-9a-z_]+))?(?:[(](?P<attrib>[ a-zA-Z0-9;:,]*)[)])?`); + static block_quotemarks_quote_open = ctRegex!(`^'{3} quote(?:[.](?P<lang>[a-z][0-9a-z_]+))?(?:[(](?P<attrib>[ a-zA-Z0-9;:,]*)[)])?`); + static block_quotemarks_table_open = ctRegex!(`^'{3} table(?:[(](?P<attrib>[ a-zA-Z0-9;:,]*)[)])?`); // ctRegex!("^'{3} table(?:\(.*?\))?"); + static block_quotemarks_close = ctRegex!(`^('{3})$`,"m"); /+ blocked markup curly +/ static block_curly_code_open = ctRegex!(`^(?:code(?:[.](?P<syntax>[a-z][0-9a-z_]+))?(?:[(](?P<attrib>[ a-zA-Z0-9;:,]*)[)])?[{][ ]*$)`); static block_curly_code_close = ctRegex!(`^([}]code)`); @@ -140,16 +148,16 @@ static template spineRgxIn() { static table_col_separator_nl = ctRegex!(`[┊]$`, "mg"); /+ inline markup footnotes endnotes +/ static inline_notes_curly_gen = ctRegex!(`~\{.+?\}~`, "m"); - static inline_notes_curly = ctRegex!(`~\{\s*(.+?)\}~`, "mg"); - static inline_notes_curly_sp_asterisk = ctRegex!(`~\{[*]+\s+(.+?)\}~`, "m"); - static inline_notes_curly_sp_plus = ctRegex!(`~\{[+]+\s+(.+?)\}~`, "m"); + static inline_notes_curly = ctRegex!(`~\{\s*(.+?)\s*\}~`, "mg"); + static inline_notes_curly_sp_asterisk = ctRegex!(`~\{[*]+\s+(.+?)\s*\}~`, "m"); + static inline_notes_curly_sp_plus = ctRegex!(`~\{[+]+\s+(.+?)\s*\}~`, "m"); static note_ref = ctRegex!(`^\S+?noteref_(?P<ref>[0-9]+)`, "mg"); // {^{73.}^}#noteref_73 static smid_inline_url_generic = ctRegex!(`(?:^|[}(\[ ])(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)[a-zA-Z0-9_#]`, "mg"); static smid_inline_url = ctRegex!(`((?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)[a-zA-Z0-9_]\S*)`, "mg"); static smid_inline_link_naked_url = ctRegex!(`(?P<pre>^|[ (\[])(?P<link>(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤)\S+?)(?=[.,;:?!'"]?([ )\]]|$))`, "mg"); static smid_inline_link_markup_regular = ctRegex!(`(?P<pre>^|[ (\[])\{\s*(?P<content>.+?)\s*\}(?P<link>(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+?)(?=[;:!,?.]?([ )\]]|$))`, "mg"); - static smid_inline_link_endnote_url_helper_punctuated = ctRegex!(`\{~\^\s+(?P<content>.+?)\}(?P<link>(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+?)(?=[.,;:?!]?([ ]|$))`, "mg"); - static smid_inline_link_endnote_url_helper = ctRegex!(`\{~\^\s+(?P<content>.+?)\}(?P<link>(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+)`, "mg"); + static smid_inline_link_endnote_url_helper_punctuated = ctRegex!(`\{~\^\s+(?P<content>.+?)\s*\}(?P<link>(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+?)(?=[.,;:?!]?([ ]|$))`, "mg"); + static smid_inline_link_endnote_url_helper = ctRegex!(`\{~\^\s+(?P<content>.+?)\s*\}(?P<link>(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+)`, "mg"); static image = ctRegex!(`([a-zA-Z0-9._-]+?\.(?:png|gif|jpg))`, "mg"); static smid_image = ctRegex!(`(?P<pre>(?:^|[ ])[{┥](?:~\^\s+|\s*))(?P<image>[a-zA-Z0-9._-]+?\.(?:png|gif|jpg))(?P<post>(?:.*?)\s*[}┝](?:image|┤.*?├|(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+?)(?=[;:!,?.]?([ )\]]|$)))`, "mg"); static smid_image_generic = ctRegex!(`(?:^|[ ])[{┥](?:~\^\s+|\s*)\S+\.(?:png|gif|jpg).*?[}┝](?:image|┤.*?├|(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+?)(?=[;:!,?.]?([ )\]]|$))`, "mg"); @@ -213,9 +221,9 @@ static template spineRgxIn() { static br_empty_line = ctRegex!(`\n[ ]*\n`, "mg"); static br_linebreaks_newlines = ctRegex!(`[\n┘┙]`, "mg"); static br_linebreaks = ctRegex!(`[┘┙]`, "mg"); - static br_line = ctRegex!(`┘`, "mg"); - static br_line_inline = ctRegex!(`┙`, "mg"); - static br_line_spaced = ctRegex!(`┚`, "mg"); + static br_line = ctRegex!(`\s*┘\s*`, "mg"); + static br_line_inline = ctRegex!(`\s*┙\s*`, "mg"); + static br_line_spaced = ctRegex!(`\s*┚\s*`, "mg"); /+ inline markup footnotes endnotes +/ static inline_notes_al = ctRegex!(`【(?:[*+]\s+|\s*)(.+?)】`, "mg"); static inline_notes_al_special = ctRegex!(`【(?:[*+]\s+)(.+?)】`, "mg"); // TODO remove match when special footnotes are implemented @@ -228,6 +236,7 @@ static template spineRgxIn() { static inline_al_delimiter_open_symbol_star = ctRegex!(`【[*]\s`, "m"); static inline_al_delimiter_open_symbol_plus = ctRegex!(`【[+]\s`, "m"); static inline_text_and_note_al_ = ctRegex!(`(.+?(?:【[*+]*\s+.+?】|.+))`, "mg"); + static endnote_section_note = ctRegex!(`┥\s*⑆\^┨(?P<notenumber>\d+)\.┣\^┝┤(?P<link>¤?.+?)├.+`, "mg"); /+ inline markup links +/ static inline_image = ctRegex!(`(?P<pre>┥)☼(?P<imginf>(?P<img>[a-zA-Z0-9._-]+?\.(?:jpg|gif|png)),w(?P<width>\d+)h(?P<height>\d+))\s*(?P<post>.*?┝┤.*?├)`, "mg"); static inline_image_without_dimensions = ctRegex!(`(?P<pre>┥)☼(?P<imginf>(?P<img>[a-zA-Z0-9._-]+?\.(?:jpg|gif|png)),w(?P<width>0)h(?P<height>0))\s*(?P<post>.*?┝┤.*?├)`, "mg"); @@ -242,6 +251,7 @@ static template spineRgxIn() { static inline_link_seg_and_hash = ctRegex!(`┥(?P<text>.+?)┝┤(?P<link>(?P<seg>[^/#├]*)#(?P<hash>.+?))├`, "mg"); static inline_link_clean = ctRegex!(`┤(?:.+?)├|[┥┝]`, "mg"); static inline_link_toc_to_backmatter = ctRegex!(`┤#(?P<link>endnotes|bibliography|bookindex|glossary|blurb)├`, "mg"); + static find_bookindex_ocn_link_and_comma = ctRegex!(`[, ]*┥.+?┝┤#?\S+?├`, "mg"); static url = ctRegex!(`https?://`, "mg"); static uri = ctRegex!(`(?:https?|git)://`, "mg"); static uri_identify_components = ctRegex!(`(?P<type>(?:https?|git)://)(?P<path>\S+?/)(?P<file>[^/]+)$`, "mg"); |