diff --git a/CHANGELOG.md b/CHANGELOG.md
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -1,6 +1,11 @@
Changelog for pdfname
=====================
+0.3
+---
+
+Added Unicode and HTML escaped text substitutions.
+
0.2
---
diff --git a/pdfname.cabal b/pdfname.cabal
--- a/pdfname.cabal
+++ b/pdfname.cabal
@@ -1,5 +1,5 @@
name: pdfname
-version: 0.2
+version: 0.3
license: MIT
license-file: LICENSE
author: Andrés Sicard-Ramírez with contributions by Juan Pedro Villa-Isaza.
@@ -15,7 +15,7 @@
The @pdfname@ command-line program names a PDF file using the
author, year of creation and title information extracted from the
@pdfinfo@ program.
-tested-with: GHC == 8.0.2
+tested-with: GHC == 8.2.1
extra-source-files: CHANGELOG.md
README.md
diff --git a/src/CreateFile.hs b/src/CreateFile.hs
--- a/src/CreateFile.hs
+++ b/src/CreateFile.hs
@@ -28,10 +28,9 @@
import Substitutions
( authorSubst
- , titleSubst
+ , removeFromTitle
, replace
- , replaceHTMLEntities
- , replaceHTMLSymbols
+ , replaceHTMLEscapedText
, unicodeSubst
, weirdSubst
)
@@ -55,8 +54,7 @@
. T.intercalate (T.singleton '-')
. map (T.reverse . T.takeWhile (' ' /=) . T.reverse)
. T.split (',' ==)
- . replaceHTMLSymbols
- . replaceHTMLEntities
+ . replaceHTMLEscapedText
. replace authorSubst
. replace weirdSubst
) xs
@@ -72,9 +70,8 @@
-- NB that the substitutions are not commutative.
( T.toLower
. replace unicodeSubst
- . replaceHTMLSymbols
- . replaceHTMLEntities
- . replace titleSubst
+ . replaceHTMLEscapedText
+ . removeFromTitle
. replace weirdSubst
) xs
diff --git a/src/Substitutions.hs b/src/Substitutions.hs
--- a/src/Substitutions.hs
+++ b/src/Substitutions.hs
@@ -2,46 +2,49 @@
module Substitutions
( authorSubst
- , titleSubst
+ , removeFromTitle
, replace
- , replaceHTMLEntities
- , replaceHTMLSymbols
+ , replaceHTMLEscapedText
, unicodeSubst
, weirdSubst
) where
+import Data.Char ( ord )
+
import Data.Text ( Text )
import qualified Data.Text as T
+import Text.Printf ( printf )
+
------------------------------------------------------------------------------
--- Replacements
+-- Replacements and pre-processing
-type HTMLEntityName = Text
-type HTMLEntityDec = Text
-type HTMLEntityHex = Text
-type HTMLSymbolDec = Text
-type HTMLSymbolHex = Text
+-- | Remove text from the title.
+removeFromTitle ∷ Text → Text
+removeFromTitle = replace $ map (\ a → (a, T.empty)) titleErase
-replaceHTMLEntities ∷ Text → Text
-replaceHTMLEntities = replace nameSubst . replace decSubst . replace hexSubst
+-- These substitutions should be done before the Unicode
+-- substitutions.
+replaceHTMLEscapedText ∷ Text → Text
+replaceHTMLEscapedText = replace $ htmlNameSubst ++ numericSubst
where
- nameSubst ∷ [(HTMLEntityName, Text)]
- nameSubst = map (\ (a, _, _, d) → (a, d)) htmlEntitySubst
+ htmlDecSubst ∷ [(Text, Text)]
+ htmlDecSubst = substHelper "&#%d;"
- decSubst ∷ [(HTMLEntityDec, Text)]
- decSubst = map (\ (_, b, _, d) → (b, d)) htmlEntitySubst
+ html4HexSubst ∷ [(Text, Text)]
+ html4HexSubst = substHelper "&#x%04X;"
- hexSubst ∷ [(HTMLEntityHex, Text)]
- hexSubst = map (\ (_, _, c, d) → (c, d)) htmlEntitySubst
+ html5HexSubst ∷ [(Text, Text)]
+ html5HexSubst = substHelper "&#x%05X;"
-replaceHTMLSymbols ∷ Text → Text
-replaceHTMLSymbols = replace decSubst . replace hexSubst
- where
- decSubst ∷ [(HTMLSymbolDec, Text)]
- decSubst = map (\ (a, _, c) → (a, c)) htmlSymbolSubst
+ numericSubst ∷ [(Text, Text)]
+ numericSubst = htmlDecSubst ++ html4HexSubst ++ html5HexSubst
- hexSubst ∷ [(HTMLSymbolHex, Text)]
- hexSubst = map (\ (_, b, c) → (b, c)) htmlSymbolSubst
+ substHelper ∷ String → [(Text, Text)]
+ substHelper format = map (\ (_, b) → (printfHelper format b, b)) htmlNameSubst
+ where
+ printfHelper ∷ String → Text → Text
+ printfHelper pFormat t = T.pack $ printf pFormat (ord $ T.head t)
replace ∷ [(Text,Text)] → Text → Text
replace xs ys = foldl (flip (uncurry T.replace)) ys xs
@@ -49,87 +52,183 @@
------------------------------------------------------------------------------
-- HTML entities and symbols substitutions
--- Used by example for the Journal of Functional Programmning.
+noName ∷ Text
+noName = "N/A"
--- | Substitutions of HTML entities.
-htmlEntitySubst ∷ [(HTMLEntityName, HTMLEntityDec, HTMLEntityHex, Text)]
-htmlEntitySubst =
- [ ("Â", "Â", "Â", "A") -- LATIN CAPITAL LETTER A WITH CIRCUMFLEX
- , ("Ä", "Ä", "Ä", "A") -- LATIN CAPITAL LETTER A WITH DIAERESIS
- , ("Ç", "Ç", "Ç", "C") -- LATIN CAPITAL LETTER C WITH CEDILLA
- , ("Ë", "Ë", "Ë", "E") -- LATIN CAPITAL LETTER E WITH DIAERESIS
- , ("Ï", "Ï", "Ï", "I") -- LATIN CAPITAL LETTER I WITH DIAERESIS
- , ("Ö", "Ö", "Ö", "O") -- LATIN CAPITAL LETTER U WITH DIAERESIS
- , ("Ø", "Ø", "Ø", "O") -- LATIN CAPITAL LETTER O WITH STROKE
- , ("Ü", "Ü", "Ü", "U") -- LATIN CAPITAL LETTER U WITH DIAERESIS
- , ("á", "á", "á", "a") -- LATIN SMALL LETTER A WITH ACUTE
- , ("â", "â", "â", "a") -- LATIN SMALL LETTER A WITH CIRCUMFLEX
- , ("ä", "ä", "ä", "a") -- LATIN SMALL LETTER A WITH DIAERESIS
- , ("ç", "ç", "ç", "c") -- LATIN SMALL LETTER C WITH CEDILLA
- , ("é", "é", "é", "e") -- LATIN SMALL LETTER E WITH ACUTE
- , ("ë", "ë", "ë", "e") -- LATIN SMALL LETTER E WITH DIAERESIS
- , ("í", "í", "í", "i") -- LATIN SMALL LETTER I WITH ACUTE
- , ("ï", "ï", "ï", "i") -- LATIN SMALL LETTER I WITH DIAERESIS
- , ("ó", "ó", "ó", "o") -- LATIN SMALL LETTER O WITH ACUTE
- , ("ö", "ö", "ö", "o") -- LATIN SMALL LETTER O WITH DIAERESIS
- , ("ø", "ø", "ø", "o") -- LATIN SMALL LETTER O WITH STROKE
- , ("ú", "ú", "ú", "u") -- LATIN SMALL LETTER U WITH ACUTE
- , ("ü", "ü", "ü", "u") -- LATIN SMALL LETTER U WITH DIAERESIS
- , ("Ī", "Ī", "Ī", "I") -- LATIN CAPITAL LETTER I WITH MACRON
- , ("ī", "ī", "ī", "I") -- LATIN SMALL LETTER I WITH MACRON
- , ("Ś", "Ś", "Ś", "S") -- LATIN CAPITAL LETTER S WITH ACUTE
- , ("ś", "ś", "ś", "s") -- LATIN SMALL LETTER S WITH ACUTE
- , ("Ş", "Ş", "Ş", "s") -- LATIN CAPITAL LETTER S WITH CEDILLA
- , ("ş", "ş", "ş", "s") -- LATIN SMALL LETTER S WITH CEDILLA
- , ("Š", "Š", "Š", "S") -- LATIN CAPITAL LETTER S WITH CARON
- , ("š", "š", "š", "s") -- LATIN SMALL LETTER S WITH CARON
- , ("α", "α", "α", "alpha") -- GREEK SMALL LETTER ALPHA
- , ("β", "β", "β", "beta") -- GREEK SMALL LETTER BETA
- , ("γ", "γ", "γ", "gamma") -- GREEK SMALL LETTER GAMMA
- , ("δ", "δ", "δ", "delta") -- GREEK SMALL LETTER DELTA
- , ("ε", "ε", "ε", "epsilon") -- GREEK SMALL LETTER EPSILON
- , ("ζ", "ζ", "ζ", "zeta") -- GREEK SMALL LETTER ZETA
- , ("η", "η", "η", "eta") -- GREEK SMALL LETTER ETA
- , ("θ", "θ", "θ", "theta") -- GREEK SMALL LETTER THETA
- , ("ι", "ι", "ι", "iota") -- GREEK SMALL LETTER IOTA
- , ("κ", "κ", "κ", "kappa") -- GREEK SMALL LETTER KAPPA
- , ("λ", "λ", "λ", "lambda") -- GREEK SMALL LETTER LAMDA
- , ("μ", "μ", "μ", "mu") -- GREEK SMALL LETTER MU
- , ("ν", "ν", "ν", "nu") -- GREEK SMALL LETTER NU
- , ("ξ", "ξ", "ξ", "xi") -- GREEK SMALL LETTER ZI
- , ("ο", "ο", "ο", "omicron") -- GREEK SMALL LETTER OMICRON
- , ("π", "π", "π", "pi") -- GREEK SMALL LETTER PI
- , ("ρ", "ρ", "ρ", "rho") -- GREEK SMALL LETTER RHO
- , ("ς", "ς", "ς", "sigma") -- GREEK SMALL LETTER FINAL SIGMA
- , ("σ", "σ", "σ", "sigma") -- GREEK SMALL LETTER SIGMA
- , ("τ", "τ", "τ", "tau") -- GREEK SMALL LETTER TAU
- , ("υ", "υ", "υ", "upsilon") -- GREEK SMALL LETTER UPSILON
- , ("φ", "φ", "φ", "phi") -- GREEK SMALL LETTER PHI
- , ("χ", "χ", "χ", "chi") -- GREEK SMALL LETTER CHI
- , ("ψ", "ψ", "ψ", "psi") -- GREEK SMALL LETTER PSI
- , ("ω", "ω", "ω", "omega") -- GREEK SMALL LETTER OMEGA
- , ("–", "–", "–", "-") -- EN DASH
- , ("—", "—", "—", ".") -- EM DAS
- , ("‘", "‘", "‘", "") -- LEFT SINGLE QUOTATION MARK
- , ("’", "’", "’", "") -- RIGHT SINGLE QUOTATION MARK
- , ("‚", "‚", "‚", "") -- SINGLE LOW-9 QUOTATION MAR
- , ("“", "“", "“", "") -- LEFT DOUBLE QUOTATION MARK
- , ("”", "”", "”", "") -- RIGHT DOUBLE QUOTATION MARK
- , ("„", "„", "„", "") -- DOUBLE LOW-9 QUOTATION MARK
- , ("†", "†", "†", "dagger") -- DAGGER
- , ("‡", "‡", "‡", "dagger-dagger") -- DOUBLE DAGGER
- , ("•", "•", "•", "") -- BULLET
- , ("…", "…", "…", "") -- HORIZONTAL ELLIPSIS
- , ("⊃", "₃", "⊃", "") -- SUPERSET OF
- ]
+-- | Substitutions of HTML-escaped text (entities and symbols).
--- | Substitutions of HTML symbols.
-htmlSymbolSubst ∷ [(HTMLSymbolDec, HTMLSymbolHex, Text)]
-htmlSymbolSubst =
- [ ("‐", "‐", "-") -- HYPHEN
- , ("‛", "‛", "") -- SINGLE HIGH-REVERSED-9 QUOTATION MARK
- , ("‟", "‟", "") -- DOUBLE HIGH-REVERSED-9 QUOTATION MARK
- , ("⌝", "⌝", "") -- TOP RIGHT CORNER
+-- We use @noName@ when the HTML symbol has not an entity name.
+htmlNameSubst ∷ [(Text, Text)]
+htmlNameSubst =
+ [ (noName, "!") -- U+0021 EXCLAMATION MARK
+ , (""", "\"") -- U+0022 QUOTATION MARK
+ , (noName, "#") -- U+0023 NUMBER SIGN
+ , (noName, "$") -- U+0024 DOLLAR SIGN
+ , (noName, "%") -- U+0025 PERCENT SIGN
+ , ("&", "&") -- U+0026 AMPERSAND
+ , (noName, "'") -- U+0027 APOSTROPHE
+ , (noName, "(") -- U+0028 LEFT PARENTHESIS
+ , (noName, ")") -- U+0029 RIGHT PARENTHESIS
+ , (noName, "*") -- U+002A ASTERISK
+ , (noName, "+") -- U+002B PLUS SIGN
+ , (noName, ",") -- U+002C COMMA
+ -- We do not substitute U+002D HYPHEN-MINUS
+ , (noName, ".") -- U+002E FULL STOP
+ , (noName, "/") -- U+002F SOLIDUS
+ , (noName, ":") -- U+003A COLON
+ , (noName, ";") -- U+003B SEMICOLON
+ , ("<", "<") -- U+003C LESS-THAN SIGN
+ , (noName, "=") -- U+003D EQUALS SIGN
+ , (">", ">") -- U+003E GREATER-THAN SIGN
+ , (noName, "?") -- U+003F QUESTION MARK
+ , (noName, "@") -- U+0040 COMMERCIAL AT
+ , ("²", "²") -- U+00B2 SUPERSCRIPT TWO
+ , ("³", "³") -- U+00B3 SUPERSCRIPT THREE
+ , ("¹", "¹") -- U+00B9 SUPERSCRIPT ONE
+ , ("À", "À") -- U+00C0
+ , ("Á", "Á") -- U+00C1
+ , ("Â", "Â") -- U+00C2
+ , ("Ã", "Ã") -- U+00C3
+ , ("Ä", "Ä") -- U+00C4
+ , ("Å", "Å") -- U+00C5
+ , ("Æ", "Æ") -- U+00C6
+ , ("Ç", "Ç") -- U+00C7
+ , ("È", "È") -- U+00C8
+ , ("É", "É") -- U+00C9
+ , ("Ê", "Ê") -- U+00CA
+ , ("Ë", "Ë") -- U+00CB
+ , ("Ì", "Ì") -- U+00CC
+ , ("Í", "Í") -- U+00CD
+ , ("Î", "Î") -- U+00CE
+ , ("Ï", "Ï") -- U+00CF
+ , ("Ð", "Ð") -- U+00D0
+ , ("Ñ", "Ñ") -- U+00D1
+ , ("Ò", "Ò") -- U+00D2
+ , ("Ó", "Ó") -- U+00D3
+ , ("Ô", "Ô") -- U+00D4
+ , ("Õ", "Õ") -- U+00D5
+ , ("Ö", "Ö") -- U+00D6
+ , ("×", "×") -- U+00D7
+ , ("Ø", "Ø") -- U+00D8
+ , ("Ù", "Ù") -- U+00D9
+ , ("Ú", "Ú") -- U+00DA
+ , ("Û", "Û") -- U+00DB
+ , ("Ü", "Ü") -- U+00DC
+ , ("Ý", "Ý") -- U+00DD
+ , ("Þ", "Þ") -- U+00DE
+ , ("ß", "ß") -- U+00DF
+ , ("à", "à") -- U+00E0
+ , ("á", "á") -- U+00E1
+ , ("â", "â") -- U+00E2
+ , ("ã", "ã") -- U+00E3
+ , ("ä", "ä") -- U+00E4
+ , ("å", "å") -- U+00E5
+ , ("æ", "æ") -- U+00E6
+ , ("ç", "ç") -- U+00E7
+ , ("è", "è") -- U+00E8
+ , ("é", "é") -- U+00E9
+ , ("ê", "ê") -- U+00EA
+ , ("ë", "ë") -- U+00EB
+ , ("ì", "ì") -- U+00EC
+ , ("í", "í") -- U+00ED
+ , ("î", "î") -- U+00EE
+ , ("ï", "ï") -- U+00EF
+ , ("ð", "ð") -- U+00F0
+ , ("ñ", "ñ") -- U+00F1
+ , ("ò", "ò") -- U+00F2
+ , ("ó", "ó") -- U+00F3
+ , ("ô", "ô") -- U+00F4
+ , ("õ", "õ") -- U+00F5
+ , ("ö", "ö") -- U+00F6
+ , ("÷", "÷") -- U+00F7
+ , ("ø", "ø") -- U+00F8
+ , ("ù", "ù") -- U+00F9
+ , ("ú", "ú") -- U+00FA
+ , ("û", "û") -- U+00FB
+ , ("ü", "ü") -- U+00FC
+ , ("ý", "ý") -- U+00FD
+ , ("þ", "þ") -- U+00FE
+ , ("ÿ", "ÿ") -- U+00FF
+ , (noName, "ĕ") -- U+0115
+ , ("Ī", "Ī") -- U+012A
+ , ("ī", "ī") -- U+012B
+ , (noName, "ĭ") -- U+012D
+ , ("Ś", "Ś") -- U+015A
+ , ("ś", "ś") -- U+015B
+ , ("Ş", "Ş") -- U+015E
+ , ("ş", "ş") -- U+015F
+ , ("Š", "Š") -- U+0160
+ , ("š", "š") -- U+0161
+ , ("ŭ", "ŭ") -- U+016D
+ , ("Α", "Α") -- U+0391
+ , ("Β", "Β") -- U+0392
+ , ("Γ", "Γ") -- U+0393
+ , ("Δ", "Δ") -- U+0394
+ , ("Ε", "Ε") -- U+0395
+ , ("Ζ", "Ζ") -- U+0396
+ , ("Η", "Η") -- U+0397
+ , ("Θ", "θ") -- U+0398
+ , ("Ι", "Ι") -- U+0399
+ , ("Κ", "Κ") -- U+03AA
+ , ("Λ", "Λ") -- U+03AB
+ , ("Μ", "Μ") -- U+03AC
+ , ("Ν", "Ν") -- U+03AD
+ , ("Ξ", "Ξ") -- U+03AE
+ , ("Ο", "Ο") -- U+03AF
+ , ("Π", "Π") -- U+03A0
+ , ("Ρ", "Ρ") -- U+03A1
+ -- The code U+03A2 is reserved.
+ , ("Σ", "Σ") -- U+03A3
+ , ("Τ", "Τ") -- U+03A4
+ , ("Υ", "Υ") -- U+03A5
+ , ("Φ", "Φ") -- U+03A6
+ , ("Χ", "Χ") -- U+03A7
+ , ("Ψ", "Ψ") -- U+03A8
+ , ("Ω", "Ω") -- U+03A9
+ , ("α", "α") -- U+03B1
+ , ("β", "β") -- U+03B2
+ , ("γ", "γ") -- U+03B3
+ , ("δ", "δ") -- U+03B4
+ , ("ε", "ε") -- U+03B5
+ , ("ζ", "ζ") -- U+03B6
+ , ("η", "η") -- U+03B7
+ , ("θ", "θ") -- U+03B8
+ , ("ι", "ι") -- U+03B9
+ , ("κ", "κ") -- U+03BA
+ , ("λ", "λ") -- U+03BB
+ , ("μ", "μ") -- U+03BC
+ , ("ν", "ν") -- U+03BD
+ , ("ξ", "ξ") -- U+03BE
+ , ("ο", "ο") -- U+03BF
+ , ("π", "π") -- U+03C0
+ , ("ρ", "ρ") -- U+03C1
+ , ("ς", "ς") -- U+03C2
+ , ("σ", "σ") -- U+03C3
+ , ("τ", "τ") -- U+03C4
+ , ("υ", "υ") -- U+03C5
+ , ("φ", "φ") -- U+03C6
+ , ("χ", "χ") -- U+03C7
+ , ("ψ", "ψ") -- U+03C8
+ , ("ω", "ω") -- U+03C9
+ , (noName, "‐") -- U+2010
+ , ("–", "–") -- U+2013
+ , ("—", "—") -- U+2014
+ , ("‘", "‘") -- U+2018
+ , ("’", "’") -- U+2019
+ , ("‚", "‚") -- U+201A
+ , (noName, "‛") -- U+201B
+ , ("“", "“") -- U+201C
+ , ("”", "”") -- U+201D
+ , ("„", "„") -- U+201E
+ , (noName, "‟") -- U+201F
+ , ("†", "†") -- U+2020
+ , ("‡", "‡") -- U+2021
+ , ("•", "•") -- U+2022
+ , ("…", "…") -- U+2026
+ , ("∀", "∀") -- U+2200
+ , ("∃", "∃") -- U+2203
+ , ("⊃", "⊃") -- U+2283
+ , (noName, "⌝") -- U+231D
]
------------------------------------------------------------------------------
@@ -155,7 +254,8 @@
, ("*", "") -- U+002A ASTERISK
, ("+", "") -- U+002B PLUS SIGN
, (",", "") -- U+002C COMMA
- , (".", "") -- U+002D FULL STOP
+ -- We do not substitute U+002D HYPHEN-MINUS
+ , (".", "") -- U+002E FULL STOP
, ("/", "") -- U+002F SOLIDUS
, (":", "") -- U+003A COLON
, (";", "") -- U+003B SEMICOLON
@@ -191,10 +291,13 @@
, ("É", "E") -- U+00C9 LATIN CAPITAL LETTER E WITH ACUTE
, ("Ê", "E") -- U+00CA LATIN CAPITAL LETTER E WITH CIRCUMFLEX
, ("Ë", "E") -- U+00CB LATIN CAPITAL LETTER E WITH DIAERESIS
+ , ("Ì", "I") -- U+00CC LATIN CAPITAL LETTER I WITH GRAVE
, ("Í", "I") -- U+00CD LATIN CAPITAL LETTER I WITH ACUTE
, ("Î", "I") -- U+00CE LATIN CAPITAL LETTER I WITH CIRCUMFLEX
, ("Ï", "I") -- U+00CF LATIN CAPITAL LETTER I WITH DIAERESIS
+ , ("Ð", "ETH") -- U+00D0 LATIN CAPITAL LETTER ETH
, ("Ñ", "N") -- U+00D1 LATIN CAPITAL LETTER N WITH TILDE
+ , ("Ò", "O") -- U+00D2 LATIN CAPITAL LETTER O WITH GRAVE
, ("Ó", "O") -- U+00D3 LATIN CAPITAL LETTER O WITH ACUTE
, ("Ô", "O") -- U+00D4 LATIN CAPITAL LETTER O WITH CIRCUMFLEX
, ("Õ", "O") -- U+00D5 LATIN CAPITAL LETTER O WITH TILDE
@@ -206,6 +309,7 @@
, ("Û", "U") -- U+00DB LATIN CAPITAL LETTER U WITH CIRCUMFLEX
, ("Ü", "U") -- U+00DC LATIN CAPITAL LETTER U WITH DIAERESIS
, ("Ý", "Y") -- U+00DD LATIN CAPITAL LETTER Y WITH ACUTE
+ , ("Þ", "THORN") -- U+00DE LATIN CAPITAL LETTER THORN
, ("ß", "ss") -- U+00DF LATIN SMALL LETTER SHARP S
, ("à", "a") -- U+00E0 LATIN SMALL LETTER A WITH GRAVE
, ("á", "a") -- U+00E1 LATIN SMALL LETTER A WITH ACUTE
@@ -219,22 +323,25 @@
, ("é", "e") -- U+00E9 LATIN SMALL LETTER E WITH ACUTE
, ("ê", "e") -- U+00EA LATIN SMALL LETTER E WITH CIRCUMFLEX
, ("ë", "e") -- U+00EB LATIN SMALL LETTER E WITH DIAERESIS
+ , ("ì", "i") -- U+00EC LATIN SMALL LETTER I WITH GRAVE
, ("í", "i") -- U+00ED LATIN SMALL LETTER I WITH ACUTE
, ("î", "i") -- U+00EE LATIN SMALL LETTER I WITH CIRCUMFLEX
, ("ï", "i") -- U+00EF LATIN SMALL LETTER I WITH DIAERESIS
+ , ("ð", "eth") -- U+00F0 LATIN SMALL LETTER ETH
, ("ñ", "n") -- U+00F1 LATIN SMALL LETTER N WITH TILDE
, ("ò", "o") -- U+00F2 LATIN SMALL LETTER O WITH GRAVE
, ("ó", "o") -- U+00F3 LATIN SMALL LETTER O WITH ACUTE
, ("ô", "o") -- U+00F4 LATIN SMALL LETTER O WITH CIRCUMFLEX
, ("õ", "o") -- U+00F5 LATIN SMALL LETTER O WITH TILDE
, ("ö", "o") -- U+00F6 LATIN SMALL LETTER O WITH DIAERESIS
+ , ("÷", "") -- U+00F7 DIVISION SIGN
, ("ø", "o") -- U+00F8 LATIN SMALL LETTER O WITH STROKE
, ("ù", "u") -- U+00F9 LATIN SMALL LETTER U WITH GRAVE
, ("ú", "u") -- U+00FA LATIN SMALL LETTER U WITH ACUTE
, ("û", "u") -- U+00FB LATIN SMALL LETTER U WITH CIRCUMFLEX
, ("ü", "u") -- U+00FC LATIN SMALL LETTER U WITH DIAERESIS
, ("ý", "y") -- U+00FD LATIN SMALL LETTER Y WITH ACUTE
- , ("þ", "t") -- U+00FE LATIN SMALL LETTER THORN
+ , ("þ", "thorn") -- U+00FE LATIN SMALL LETTER THORN
, ("ÿ", "y") -- U+00FF LATIN SMALL LETTER Y WITH DIAERESIS
, ("Ā", "A") -- U+0100 LATIN CAPITAL LETTER A WITH MACRON
, ("ā", "a") -- U+0101 LATIN SMALL LETTER A WITH MACRON
@@ -250,6 +357,7 @@
, ("ď", "d") -- U+010F LATIN SMALL LETTER D WITH CARON
, ("Ē", "E") -- U+0112 LATIN CAPITAL LETTER E WITH MACRON
, ("ē", "e") -- U+0113 LATIN SMALL LETTER E WITH MACRON
+ , ("ĕ", "e") -- U+0115 LATIN SMALL LETTER E WITH BREVE
, ("Ę", "e") -- U+0118 LATIN CAPITAL LETTER E WITH OGONEK
, ("ę", "e") -- U+0119 LATIN SMALL LETTER E WITH OGONEK
, ("Ě", "E") -- U+011A LATIN CAPITAL LETTER E WITH CARON
@@ -260,6 +368,7 @@
, ("ģ", "g") -- U+0123 LATIN SMALL LETTER G WITH CEDILLA
, ("Ī", "I") -- U+012A LATIN CAPITAL LETTER I WITH MACRON
, ("ī", "I") -- U+012B LATIN SMALL LETTER I WITH MACRON
+ , ("ĭ", "i") -- U+012D LATIN SMALL LETTER I WITH BREVE
, ("İ", "I") -- U+0130 LATIN CAPITAL LETTER I WITH DOT ABOVE
, ("Ķ", "K") -- U+0136 LATIN CAPITAL LETTER K WITH CEDILLA
, ("ķ", "k") -- U+0137 LATIN SMALL LETTER K WITH CEDILLA
@@ -288,6 +397,7 @@
, ("Ť", "T") -- U+0164 LATIN CAPITAL LETTER T WITH CARON
, ("ť", "t") -- U+0165 LATIN SMALL LETTER T WITH CARON
, ("Ū", "U") -- U+016A LATIN CAPITAL LETTER U WITH MACRON
+ , ("ŭ", "u") -- U+016D LATIN SMALL LETTER U WITH BREVE
, ("ū", "u") -- U+016B LATIN SMALL LETTER U WITH MACRON
, ("Ů", "U") -- U+016E LATIN CAPITAL LETTER U WITH RING ABOVE
, ("ů", "u") -- U+016F LATIN SMALL LETTER U WITH RING ABOVE
@@ -306,6 +416,30 @@
, ("Ț", "T") -- U+021A LATIN CAPITAL LETTER T WITH COMMA BELOW
, ("ț", "t") -- U+021B LATIN SMALL LETTER T WITH COMMA BELOW
, ("ə", "a") -- U+0259 LATIN SMALL LETTER SCHWA
+ , ("Α", "Alpha") -- U+0391 GREEK CAPITAL LETTER ALPHA
+ , ("Β", "Beta") -- U+0392 GREEK CAPITAL LETTER BETA
+ , ("Γ", "Gamma") -- U+0393 GREEK CAPITAL LETTER GAMMA
+ , ("Δ", "Delta") -- U+0394 GREEK CAPITAL LETTER DELTA
+ , ("Ε", "Epsilon") -- U+0395 GREEK CAPITAL LETTER EPSILON
+ , ("Ζ", "Zeta") -- U+0396 GREEK CAPITAL LETTER ZETA
+ , ("Η", "Eta") -- U+0397 GREEK CAPITAL LETTER ETA
+ , ("Θ", "Theta") -- U+0398 GREEK CAPITAL LETTER THETA
+ , ("Ι", "Iota") -- U+0399 GREEK CAPITAL LETTER IOTA
+ , ("Κ", "Kappa") -- U+039A GREEK CAPITAL LETTER KAPPA
+ , ("Λ", "Lambda") -- U+039B GREEK CAPITAL LETTER LAMDA
+ , ("Μ", "Mu") -- U+039C GREEK CAPITAL LETTER MU
+ , ("Ν", "Nu") -- U+039D GREEK CAPITAL LETTER NU
+ , ("Ξ", "Xi") -- U+039E GREEK CAPITAL LETTER ZI
+ , ("Ο", "Omicron") -- U+039F GREEK CAPITAL LETTER OMICRON
+ , ("Π", "Pi") -- U+03A0 GREEK CAPITAL LETTER PI
+ , ("Ρ", "Rho") -- U+03A1 GREEK CAPITAL LETTER RHO
+ -- The code U+03A2 is reserved.
+ , ("Σ", "Sigma") -- U+03A3 GREEK CAPITAL LETTER SIGMA
+ , ("Τ", "Tau") -- U+03A4 GREEK CAPITAL LETTER TAU
+ , ("Υ", "Upsilon") -- U+03A5 GREEK CAPITAL LETTER UPSILON
+ , ("Φ", "Phi") -- U+03A6 GREEK CAPITAL LETTER PHI
+ , ("χ", "Chi") -- U+03A7 GREEK CAPITAL LETTER CHI
+ , ("Ψ", "Psi") -- U+03A8 GREEK CAPITAL LETTER PSI
, ("Ω", "Omega") -- U+03A9 GREEK CAPITAL LETTER OMEGA
, ("α", "alpha") -- U+03B1 GREEK SMALL LETTER ALPHA
, ("β", "beta") -- U+03B2 GREEK SMALL LETTER BETA
@@ -333,7 +467,8 @@
, ("ψ", "psi") -- U+03C8 GREEK SMALL LETTER PSI
, ("ω", "omega") -- U+03C9 GREEK SMALL LETTER OMEGA
, ("ẞ", "SS") -- U+1E9E LATIN CAPITAL LETTER SHARP S
- , ("‐", "-") -- U+2210 HYPHEN
+ , (" ", "-") -- U+2009 THIN SPACE
+ , ("‐", "-") -- U+2010 HYPHEN
, ("–", "-") -- U+2013 EN DASH
, ("—", "-") -- U+2014 EM DASH
, ("‘", "") -- U+2018 LEFT SINGLE QUOTATION MARK
@@ -348,9 +483,30 @@
, ("‡", "") -- U+2021 DOUBLE DAGGER
, ("•", "") -- U+2022 BULLET
, ("…", "") -- U+2026 HORIZONTAL ELLIPSIS
+ , ("⁰", "0") -- U+2070 SUPERSCRIPT ZERO
+ , ("ⁱ", "i") -- U+2071 SUPERSCRIPT LATIN SMALL LETTER I
+ -- The codes U+2072 and U+2073 are reserved.
+ , ("⁴", "4") -- U+2074 SUPERSCRIPT FOUR
+ , ("⁵", "5") -- U+2075 SUPERSCRIPT FIVE
+ , ("⁶", "6") -- U+2076 SUPERSCRIPT SIX
+ , ("⁷", "7") -- U+2077 SUPERSCRIPT SEVEN
+ , ("⁸", "8") -- U+2078 SUPERSCRIPT EIGHT
+ , ("⁹", "9") -- U+2079 SUPERSCRIPT NINE
+ , ("₀", "0") -- U+2080 SUBSCRIPT ZERO
+ , ("₁", "1") -- U+2081 SUBSCRIPT ONE
+ , ("₂", "2") -- U+2082 SUBSCRIPT TWO
+ , ("₃", "3") -- U+2083 SUBSCRIPT THREE
+ , ("₄", "4") -- U+2074 SUBSCRIPT FOUR
+ , ("₅", "5") -- U+2075 SUBSCRIPT FIVE
+ , ("₆", "6") -- U+2076 SUBSCRIPT SIX
+ , ("₇", "7") -- U+2077 SUBSCRIPT SEVEN
+ , ("₈", "8") -- U+2078 SUBSCRIPT EIGHT
+ , ("₉", "9") -- U+2079 SUBSCRIPT NINE
, ("™", "") -- U+2122 TRADE MARK SIGN
, ("ℵ", "aleph") -- U+2135 ALEF SYMBOL
, ("ℶ", "beth") -- U+2136 BET SYMBOL
+ , ("∀", "for-all") -- U+2200 FOR ALL
+ , ("∃", "exists") -- U+2203 THERE EXISTS
, ("⊃", "") -- U+2283 SUPERSET OF
, ("⌝", "") -- U+231D TOP RIGHT CORNER
, ("�", "") -- U+FFFD REPLACEMENT CHARACTER
@@ -366,42 +522,6 @@
, (" and", ",")
]
--- | Title substitutions.
-
--- These substitutions should be done before the substitutions of the
--- HTML entities and converting to lower case.
-titleSubst ∷ [(Text,Text)]
-titleSubst =
- [ ("P ", "P")
- , ("0 ", "0")
- , ("C", "C")
- , ("CC", "CC")
- , ("I ", "I")
- , ("J", "J")
- , ("Modus ponens", "Modus ponens")
- , ("P ", "P")
- , ("really ", "really")
- , ("S-P", "S-P")
- , ("3", "3")
- -- The whitespace around `+` is not the standard one.
- -- TODO (2017-07-04): Added test case.
- , (" + ", "plus")
- , ("ω", "omega")
- , ("$\\alpha$", "alpha")
- , ("$\\beta$", "beta")
- , ("$\\gamma$", "gamma")
- , ("$\\epsilon$", "epsilon")
- , ("$\\eta$", "eta")
- , ("$\\lambda$", "lambda")
- , ("$\\pi$", "pi")
- , ("$\\omega$", "omega")
- , ("{\\sc Coq}", "Coq")
- , ("{\\sf Haskell}:", "Haskell")
- , ("{\\sc QuickSpec}:", "QuickSpec")
- , ("{\\sc QuodLibet}!", "QuodLibet")
- , ("{\\sc Vampire}", "Vampire")
- ]
-
------------------------------------------------------------------------------
-- Weird substitutions.
@@ -417,8 +537,25 @@
-- We erase `ö` because it follows an `o` in the examples we know.
, ("ö", "") -- U+00C3 and U+00B6 (LATIN SMALL LETTER O WITH DIAERESIS)
, ("Å›", "s") -- U+00C5 and U+203A (LATIN CAPITAL LETTER S WITH ACUTE)
- -- TODO (2017-07-20): Five hex numbers
- , ("‐", "-") -- U+2010 HYPHEN
-- TODO (2017-07-17): Missing `;`.
, ("’", "") -- U+2019 RIGHT SINGLE QUOTATION MARK
+ ]
+
+------------------------------------------------------------------------------
+-- Title pre-processing
+
+titleErase ∷ [Text]
+titleErase =
+ [ "" -- opening
+ , "" -- opening
+ , "" -- closing
+ , "" -- opening
+ , "" -- closing
+ , "" -- opening
+ , "" -- closing
+ , "$\\" -- opening
+ , "$" -- closing
+ , "{\\sc " -- opening
+ , "{\\sf " -- opening
+ , "}" -- closing
]