# SPDX-FileCopyrightText: 2025 Dominick Johnson # SPDX-License-Identifier: MPL-2.0 # This mapping comes from appendix D of the PDF 2.0 spec. However, I doubt it is # exhaustive. It may be that an exhaustive list is not possible. from __future__ import annotations CHARNAMES_TO_UNICODE = { #### Table D.2 #### '/A': 'A', '/OE': 'Œ', '/AE': 'Æ', '/Oacute': 'Ó', '/Aacute': 'Á', '/Ocircumflex': 'Ô', '/Acircumflex': 'Â', '/Odieresis': 'Ö', '/Adieresis': 'Ä', '/Ograve': 'Ò', '/Agrave': 'À', '/Oslash': 'Ø', '/Aring': 'Å', '/Otilde': 'Õ', '/Atilde': 'Ã', '/P': 'P', '/B': 'B', '/Q': 'Q', '/C': 'C', '/R': 'R', '/Ccedilla': 'Ç', '/S': 'S', '/D': 'D', '/Scaron': 'Š', '/E': 'E', '/T': 'T', '/Eacute': 'É', '/Thorn': 'Þ', '/Ecircumflex': 'Ê', '/U': 'U', '/Edieresis': 'Ë', '/Uacute': 'Ú', '/Egrave': 'È', '/Ucircumflex': 'Û', '/Eth': 'Ð', '/Udieresis': 'Ü', '/Euro': '€', '/Ugrave': 'Ù', '/F': 'F', '/V': 'V', '/G': 'G', '/W': 'W', '/H': 'H', '/X': 'X', '/I': 'I', '/Y': 'Y', '/Iacute': 'Í', '/Yacute': 'Ý', '/Icircumflex': 'Î', '/Ydieresis': 'Ÿ', '/Idieresis': 'Ï', '/Z': 'Z', '/Igrave': 'Ì', '/Zcaron': 'Ž', '/J': 'J', '/a': 'a', '/K': 'K', '/aacute': 'á', '/L': 'L', '/acircumflex': 'â', '/Lslash': 'Ł', '/acute': '´', '/M': 'M', '/adieresis': 'ä', '/N': 'N', '/ae': 'æ', '/Ntilde': 'Ñ', '/agrave': 'à', '/O': 'O', '/ampersand': '&', '/aring': 'å', '/eth': 'ð', '/asciicircum': '^', '/exclam': '!', '/asciitilde': '~', '/exclamdown': '¡', '/asterisk': '*', '/f': 'f', '/at': '@', '/fi': 'fi', '/atilde': 'ã', '/five': '5', '/b': 'b', '/fl': 'fl', '/backslash': '\\', '/florin': 'ƒ', '/bar': '|', '/four': '4', '/braceleft': '{', '/fraction': '⁄', '/braceright': '}', '/g': 'g', '/bracketleft': '[', '/germandbls': 'ß', '/bracketright': ']', '/grave': '`', '/breve': '˘', '/greater': '>', '/brokenbar': '¦', '/guillemotleft': '«', '/bullet': '•', '/guillemotright': '»', '/c': 'c', '/guilsinglleft': '‹', '/caron': 'ˇ', '/guilsinglright': '›', '/ccedilla': 'ç', '/h': 'h', '/cedilla': '¸', '/hungarumlaut': '˝', '/cent': '¢', '/hyphen': '-', '/circumflex': 'ˆ', '/i': 'i', '/colon': ':', '/iacute': 'í', '/comma': ',', '/icircumflex': 'î', '/copyright': '©', '/idieresis': 'ï', '/currency': '¤', '/igrave': 'ì', '/d': 'd', '/j': 'j', '/dagger': '†', '/k': 'k', '/daggerdbl': '‡', '/l': 'l', '/degree': '°', '/less': '<', '/dieresis': '¨', '/logicalnot': '¬', '/divide': '÷', '/lslash': 'ł', '/dollar': '$', '/m': 'm', '/dotaccent': '˙', '/macron': '¯', '/dotlessi': 'ı', '/minus': '−', '/e': 'e', '/mu': 'μ', '/eacute': 'é', '/multiply': '×', '/ecircumflex': 'ê', '/n': 'n', '/edieresis': 'ë', '/nine': '9', '/egrave': 'è', '/ntilde': 'ñ', '/eight': '8', '/numbersign': '#', '/ellipsis': '…', '/o': 'o', '/emdash': '—', '/oacute': 'ó', '/endash': '–', '/ocircumflex': 'ô', '/equal': '=', '/odieresis': 'ö', '/oe': 'œ', '/s': 's', '/ogonek': '˛', '/scaron': 'š', '/ograve': 'ò', '/section': '§', '/one': '1', '/semicolon': ';', '/onehalf': '½', '/seven': '7', '/onequarter': '¼', '/six': '6', '/onesuperior': '¹', '/slash': '/', '/ordfeminine': 'ª', '/space': ' ', '/ordmasculine': 'º', '/sterling': '£', '/oslash': 'ø', '/t': 't', '/otilde': 'õ', '/thorn': 'þ', '/p': 'p', '/three': '3', '/paragraph': '¶', '/threequarters': '¾', '/parenleft': '(', '/threesuperior': '³', '/parenright': ')', '/tilde': '˜', '/percent': '%', '/trademark': '™', '/period': '.', '/two': '2', '/periodcentered': '·', '/twosuperior': '²', '/perthousand': '‰', '/u': 'u', '/plus': '+', '/uacute': 'ú', '/plusminus': '±', '/ucircumflex': 'û', '/q': 'q', '/udieresis': 'ü', '/question': '?', '/ugrave': 'ù', '/questiondown': '¿', '/underscore': '_', '/quotedbl': '"', '/v': 'v', '/quotedblbase': '„', '/w': 'w', '/quotedblleft': '“', '/x': 'x', '/quotedblright': '”', '/y': 'y', '/quoteleft': '‘', '/yacute': 'ý', '/quoteright': '’', '/ydieresis': 'ÿ', '/quotesinglbase': '‚', '/yen': '¥', '/quotesingle': "'", '/z': 'z', '/r': 'r', '/zcaron': 'ž', '/registered': '®', '/zero': '0', '/ring': '˚', #### Table D.4 #### # Much of this table is commented out on account of (as far as I can tell) not # mapping to real unicode values. The values below seem to be substitutions I got # when copy/pasting from the spec. Such values might still be useful if our intent # was to convert PDF to text, but currently our intent is only to do the inverse. # '/AEsmall': 'ᴁ', # '/Jsmall': 'ᴊ', # '/Aacutesmall': 'Á', # '/Ksmall': 'ᴋ', # '/Acircumflexsmall': 'Â', # '/Lslashsmall': 'ᴌ', # '/Acutesmall': '´', # '/Lsmall': 'ʟ', # '/Adieresissmall': 'Ä', # '/Macronsmall': '¯', # '/Agravesmall': 'À', # '/Msmall': 'ᴍ', # '/Aringsmall': 'Å', # '/Nsmall': 'ɴ', # '/Asmall': 'ᴀ', # '/Ntildesmall': 'Ñ', # '/Atildesmall': 'Ã', # '/OEsmall': 'ɶ', # '/Brevesmall': '˘', # '/Oacutesmall': 'Ó', # '/Bsmall': 'ʙ', # '/Ocircumflexsmall': 'Ô', # '/Caronsmall': 'ˇ', # '/Odieresissmall': 'Ö', # '/Ccedillasmall': 'Ç', # '/Ogoneksmall': '˛', # '/Cedillasmall': '¸', # '/Ogravesmall': 'ò', # '/Circumflexsmall': 'ˆ', # '/Oslashsmall': 'ø', # '/Csmall': 'ᴄ', # '/Osmall': 'ᴏ', # '/Dieresissmall': '¨', # '/Otildesmall': 'Õ', # '/Dotaccentsmall': '˙', # '/Psmall': 'ᴘ', # '/Dsmall': 'ᴅ', # '/Qsmall': 'Q', # '/Eacutesmall': 'É', # '/Ringsmall': '˚', # '/Ecircumflexsmall': 'Ê', # '/Rsmall': 'R', # '/Edieresissmall': 'Ë', # '/Scaronsmall': 'Š', # '/Egravesmall': 'È', # '/Ssmall': 'S', # '/Esmall': 'ᴇ', # '/Thornsmall': 'þ', # '/Ethsmall': 'ᴆ', # '/Tildesmall': '˜', # '/Fsmall': 'F', # '/Tsmall': 'ᴛ', # '/Gravesmall': '`', # '/Uacutesmall': 'Ú', # '/Gsmall': 'G', # '/Ucircumflexsmall': 'Û', # '/Hsmall': 'ʜ', # '/Udieresissmall': 'Ü', # '/Hungarumlautsmall': '˝', # '/Ugravesmall': 'Ù', # '/Iacutesmall': 'Í', # '/Usmall': 'ᴜ', # '/Icircumflexsmall': 'Î', # '/Vsmall': 'ᴠ', # '/Idieresissmall': 'Ï', # '/Wsmall': 'ᴡ', # '/Igravesmall': 'Ì', # '/Xsmall': 'X', # '/Ismall': 'I', # '/Yacutesmall': 'Ý', # '/Ydieresissmall': 'Ÿ', # '/fouroldstyle': '4', # '/Ysmall': 'Y', '/foursuperior': '⁴', # '/Zcaronsmall': 'Ž', # '/Zsmall': 'z', # '/ampersandsmall': '&', # '/hypheninferior': '-', # '/asuperior': 'a', # '/hyphensuperior': '-', # '/bsuperior': 'b', # '/isuperior': 'i', # '/centinferior': '¢', # '/lsuperior': 'l', # '/centoldstyle': '¢', # '/msuperior': 'm', # '/centsuperior': '¢', '/nineinferior': '₉', # '/nineoldstyle': '9', '/colonmonetary': '₡', '/ninesuperior': '⁹', '/nsuperior': 'ⁿ', # '/commainferior': ',', '/onedotenleader': '․', # '/commasuperior': ',', '/oneeighth': '⅛', # '/dollarinferior': '$', '/onefitted': '1', # '/dollaroldstyle': '$', # '/dollarsuperior': '$', '/oneinferior': '₁', # '/dsuperior': 'd', # '/oneoldstyle': '1', '/eightinferior': '₈', # '/eightoldstyle': '8', '/eightsuperior': '⁸', '/onethird': '⅓', # '/esuperior': 'e', # '/osuperior': 'O', # '/exclamdownsmall': '¡', '/parenleftinferior': '₍', # '/exclamsmall': '!', '/parenleftsuperior': '⁽', '/ff': 'ff', '/parenrightinferior': '₎', '/ffi': 'ffi', '/parenrightsuperior': '⁾', '/ffl': 'ffl', # '/periodinferior': '.', '/figuredash': '‒', # '/periodsuperior': '.', '/fiveeighths': '⅝', # '/questiondownsmall': '¿', '/fiveinferior': '₅', # '/questionsmall': '?', # '/fiveoldstyle': '5', # '/rsuperior': 'r', '/fivesuperior': '⁵', # '/rupiah': 'Rp', '/fourinferior': '₄', '/seveneighths': '⅞', '/seveninferior': '₇', # '/threequartersemdash': '—', # '/sevenoldstyle': '7', '/sevensuperior': '⁷', # '/tsuperior': 't', '/sixinferior': '₆', '/twodotenleader': '‥', # '/sixoldstyle': '6', '/twoinferior': '₂', '/sixsuperior': '⁶', # '/twooldstyle': '2', # '/ssuperior': 'S', '/twothirds': '⅔', '/threeeighths': '⅜', '/zeroinferior': '₀', '/threeinferior': '₃', # '/zerooldstyle': '0', # '/threeoldstyle': '3', '/zerosuperior': '⁰', #### Table D.5 #### '/Alpha': 'Α', '/arrowboth': '↔', '/Beta': 'Β', '/arrowdblboth': '⇔', '/Chi': 'Χ', '/arrowdbldown': '⇓', '/Delta': 'Δ', '/arrowdblleft': '⇐', '/Epsilon': 'Ε', '/arrowdblright': '⇒', '/Eta': 'Η', '/arrowdblup': '⇑', '/arrowdown': '↓', '/Gamma': 'Γ', '/arrowhorizex': '⎯', '/Ifraktur': 'ℑ', '/arrowleft': '←', '/Iota': 'Ι', '/arrowright': '→', '/Kappa': 'Κ', '/arrowup': '↑', '/Lambda': 'Λ', '/arrowvertex': '⏐', '/Mu': 'Μ', '/asteriskmath': '∗', '/Nu': 'Ν', '/Omega': 'Ω', '/beta': 'β', '/Omicron': 'Ο', '/Phi': 'Φ', '/Pi': 'Π', '/bracelefttp': '⎧', '/Psi': 'Ψ', '/braceleftmid': '⎨', '/Rfraktur': 'ℜ', '/braceleftbt': '⎩', '/Rho': 'Ρ', '/bracerighttp': '⎫', '/Sigma': 'Σ', '/bracerightmid': '⎬', '/Tau': 'Τ', '/bracerightbt': '⎭', '/Theta': 'Θ', '/braceex': '⎪', '/Upsilon': 'Υ', '/Upsilon1': 'ϒ', '/Xi': 'Ξ', '/bracketlefttp': '⎡', '/Zeta': 'Ζ', '/bracketleftex': '⎢', '/aleph': 'ℵ', '/bracketleftbt': '⎣', '/alpha': 'α', '/bracketrighttp': '⎤', '/bracketrightex': '⎥', '/angle': '∠', '/bracketrightbt': '⎦', '/angleleft': '〈', '/angleright': '〉', '/carriagereturn': '↵', '/approxequal': '≈', '/chi': 'χ', '/circlemultiply': '⊗', '/integralbt': '⌡', '/circleplus': '⊕', '/intersection': '∩', '/club': '♣', '/iota': 'ι', '/kappa': 'κ', '/lambda': 'λ', '/congruent': '≅', '/copyrightsans': '©', '/lessequal': '≤', '/copyrightserif': '©', '/logicaland': '∧', '/delta': 'δ', '/logicalor': '∨', '/diamond': '♦', '/lozenge': '◊', '/dotmath': '⋅', '/minute': '′', '/element': '∈', '/emptyset': '∅', '/notelement': '∉', '/epsilon': 'ε', '/notequal': '≠', '/notsubset': '⊄', '/equivalence': '≡', '/nu': 'ν', '/eta': 'η', '/omega': 'ω', '/existential': '∃', '/omega1': 'ϖ', '/omicron': 'ο', '/gamma': 'γ', '/parenlefttp': '⎛', '/gradient': '∇', '/parenleftex': '⎜', '/parenleftbt': '⎝', '/greaterequal': '≥', '/parenrighttp': '⎞', '/heart': '♥', '/parenrightex': '⎟', '/infinity': '∞', '/parenrightbt': '⎠', '/integral': '∫', '/partialdiff': '∂', '/integraltp': '⌠', '/integralex': '⎮', '/perpendicular': '⊥', '/similar': '∼', '/phi': 'ϕ', '/phi1': 'φ', '/pi': 'π', '/spade': '♠', '/suchthat': '∋', '/product': 'Π', '/summation': 'Σ', '/propersubset': '⊂', '/tau': 'τ', '/propersuperset': '⊃', '/therefore': '∴', '/proportional': '∝', '/theta': 'θ', '/psi': 'ψ', '/theta1': 'ϑ', '/radical': '√', '/trademarksans': '™', '/radicalex': '⎯', '/trademarkserif': '™', '/reflexsubset': '⊆', '/reflexsuperset': '⊇', '/registersans': '®', '/union': '∪', '/registerserif': '®', '/universal': '∀', '/rho': 'ρ', '/upsilon': 'υ', '/second': '″', '/weierstrass': '℘', '/xi': 'ξ', '/sigma': 'σ', '/zeta': 'ζ', '/sigma1': 'ς', #### Table D.6 #### '/a30': '✣', '/a65': '❆', '/a109': '♠', '/a1': '✁', '/a31': '✤', '/a66': '❇', '/a120': '①', '/a2': '✂', '/a32': '✥', '/a67': '❈', '/a121': '②', '/a202': '✃', '/a33': '✦', '/a68': '❉', '/a122': '③', '/a3': '✄', '/a34': '✧', '/a69': '❊', '/a123': '④', '/a4': '☎', '/a35': '★', '/a70': '❋', '/a124': '⑤', '/a5': '✆', '/a36': '✩', '/a71': '●', '/a125': '⑥', '/a119': '✇', '/a37': '✪', '/a72': '❍', '/a126': '⑦', '/a118': '✈', '/a38': '✫', '/a73': '■', '/a127': '⑧', '/a117': '✉', '/a39': '✬', '/a74': '❏', '/a128': '⑨', '/a11': '☛', '/a40': '✭', '/a203': '❐', '/a129': '⑩', '/a12': '☞', '/a41': '✮', '/a75': '❑', '/a130': '❶', '/a13': '✌', '/a42': '✯', '/a204': '❒', '/a131': '❷', '/a14': '✍', '/a43': '✰', '/a76': '▲', '/a132': '❸', '/a15': '✎', '/a44': '✱', '/a77': '▼', '/a133': '❹', '/a16': '✏', '/a45': '✲', '/a78': '◆', '/a134': '❺', '/a105': '✐', '/a46': '✳', '/a79': '❖', '/a135': '❻', '/a17': '✑', '/a47': '✴', '/a81': '◗', '/a136': '❼', '/a18': '✒', '/a48': '✵', '/a82': '❘', '/a137': '❽', '/a19': '✓', '/a49': '✶', '/a83': '❙', '/a138': '❾', '/a20': '✔', '/a50': '✷', '/a84': '❚', '/a139': '❿', '/a21': '✕', '/a51': '✸', '/a97': '❛', '/a140': '➀', '/a22': '✖', '/a52': '✹', '/a98': '❜', '/a141': '➁', '/a23': '✗', '/a53': '✺', '/a99': '❝', '/a142': '➂', '/a24': '✘', '/a54': '✻', '/a100': '❞', '/a143': '➃', '/a25': '✙', '/a55': '✼', '/a101': '❡', '/a144': '➄', '/a26': '✚', '/a56': '✽', '/a102': '❢', '/a145': '➅', '/a27': '✛', '/a57': '✾', '/a103': '❣', '/a146': '➆', '/a28': '✜', '/a58': '✿', '/a104': '❤', '/a147': '➇', '/a6': '✝', '/a59': '❀', '/a106': '❥', '/a148': '➈', '/a7': '✞', '/a60': '❁', '/a107': '❦', '/a149': '➉', '/a8': '✟', '/a61': '❂', '/a108': '❧', '/a150': '➊', '/a9': '✠', '/a62': '❃', '/a112': '♣', '/a151': '➋', '/a10': '✡', '/a63': '❄', '/a111': '♦', '/a152': '➌', '/a29': '✢', '/a64': '❅', '/a110': '♥', '/a153': '➍', '/a154': '➎', '/a192': '➚', '/a176': '➦', '/a184': '➳', '/a155': '➏', '/a166': '➛', '/a177': '➧', '/a197': '➴', '/a156': '➐', '/a167': '➜', '/a178': '➨', '/a185': '➵', '/a157': '➑', '/a168': '➝', '/a179': '➩', '/a194': '➶', '/a158': '➒', '/a169': '➞', '/a193': '➪', '/a198': '➷', '/a159': '➓', '/a170': '➟', '/a180': '➫', '/a186': '➸', '/a160': '➔', '/a171': '➠', '/a199': '➬', '/a195': '➹', '/a161': '→', '/a172': '➡', '/a181': '➭', '/a187': '➺', '/a163': '↔', '/a173': '➢', '/a200': '➮', '/a188': '➻', '/a164': '↕', '/a162': '➣', '/a182': '➯', '/a189': '➼', '/a196': '➘', '/a174': '➤', '/a201': '➱', '/a190': '➽', '/a165': '➙', '/a175': '➥', '/a183': '➲', '/a191': '➾', }