PDF に謎の漢字が含まれるとき

gistfile1.md

PDF に謎の漢字が含まれるとき

PDF などの中にある一部の日本語の漢字が、見た目は同じだけど異なる謎の文字に変換されていることがある

• 例 1: https://www.mhlw.go.jp/content/10906000/000628667.pdf
  • 「長野」と「長崎」の「長」が、 U+9577 ではなく「⾧ (U+2FA7)」になっている
• 例 2: https://www.dpri.kyoto-u.ac.jp/news/12739/
  • 大量にある、どうしてこうなった
  • PDF ではないので何かからコピーして書いた？

この文字は 康煕部首 (Kangxi Radicals) というもので、部首としての文字である

• ＭＳ ゴシックなど Kangxi Radicals の字形がないフォントを指定すると表示できないので区別しやすい

どこから来たのか？

これらは（フォントによるが）通常の漢字と同じ字形を持っていることが多く、また U+2F00-U+2FD5 に配置されているので通常の漢字 (U+4E00 以降) より前にある。 そのため、仮想 PDF プリンタなどで変換するときに誤って参照されている可能性がある （詳しくは調べてない）

例えば、フォントに「メイリオ」を指定して 長崎県 とだけ書いた .docx ファイルを用意し、Microsoft Print to PDF を使って PDF に変換すると、「長 (U+9577)」が「⾧ (U+2FA7)」に変換されることが再現できる。 Word で直接 PDF ファイルを保存すると、変換されずに保存できる

対策

自分で PDF を作ることはあまりないが、ほかの PDF からコピーしたものに Kangxi Radicals が含まれてしまうことがある。 これを気づいて元の漢字に戻せるように textlint rule を作ってある:

• textlint-rule-no-kangxi-radicals

ここにも変換表を貼っておく:

{"⼀":"一","⼁":"丨","⼂":"丶","⼃":"丿","⼄":"乙","⼅":"亅","⼆":"二","⼇":"亠","⼈":"人","⼉":"儿","⼊":"入","⼋":"八","⼌":"冂","⼍":"冖","⼎":"冫","⼏":"几","⼐":"凵","⼑":"刀","⼒":"力","⼓":"勹","⼔":"匕","⼕":"匚","⼖":"匸","⼗":"十","⼘":"卜","⼙":"卩","⼚":"厂","⼛":"厶","⼜":"又","⼝":"口","⼞":"囗","⼟":"土","⼠":"士","⼡":"夂","⼢":"夊","⼣":"夕","⼤":"大","⼥":"女","⼦":"子","⼧":"宀","⼨":"寸","⼩":"小","⼪":"尢","⼫":"尸","⼬":"屮","⼭":"山","⼮":"巛","⼯":"工","⼰":"己","⼱":"巾","⼲":"干","⼳":"幺","⼴":"广","⼵":"廴","⼶":"廾","⼷":"弋","⼸":"弓","⼹":"彐","⼺":"彡","⼻":"彳","⼼":"心","⼽":"戈","⼾":"戸","⼿":"手","⽀":"支","⽁":"攴","⽂":"文","⽃":"斗","⽄":"斤","⽅":"方","⽆":"无","⽇":"日","⽈":"曰","⽉":"月","⽊":"木","⽋":"欠","⽌":"止","⽍":"歹","⽎":"殳","⽏":"毋","⽐":"比","⽑":"毛","⽒":"氏","⽓":"气","⽔":"水","⽕":"火","⽖":"爪","⽗":"父","⽘":"爻","⽙":"爿","⽚":"片","⽛":"牙","⽜":"牛","⽝":"犬","⽞":"玄","⽟":"玉","⽠":"瓜","⽡":"瓦","⽢":"甘","⽣":"生","⽤":"用","⽥":"田","⽦":"疋","⽧":"疒","⽨":"癶","⽩":"白","⽪":"皮","⽫":"皿","⽬":"目","⽭":"矛","⽮":"矢","⽯":"石","⽰":"示","⽱":"禸","⽲":"禾","⽳":"穴","⽴":"立","⽵":"竹","⽶":"米","⽷":"糸","⽸":"缶","⽹":"网","⽺":"羊","⽻":"羽","⽼":"老","⽽":"而","⽾":"耒","⽿":"耳","⾀":"聿","⾁":"肉","⾂":"臣","⾃":"自","⾄":"至","⾅":"臼","⾆":"舌","⾇":"舛","⾈":"舟","⾉":"艮","⾊":"色","⾋":"艸","⾌":"虍","⾍":"虫","⾎":"血","⾏":"行","⾐":"衣","⾑":"襾","⾒":"見","⾓":"角","⾔":"言","⾕":"谷","⾖":"豆","⾗":"豕","⾘":"豸","⾙":"貝","⾚":"赤","⾛":"走","⾜":"足","⾝":"身","⾞":"車","⾟":"辛","⾠":"辰","⾡":"辵","⾢":"邑","⾣":"酉","⾤":"釆","⾥":"里","⾦":"金","⾧":"長","⾨":"門","⾩":"阜","⾪":"隶","⾫":"隹","⾬":"雨","⾭":"靑","⾮":"非","⾯":"面","⾰":"革","⾱":"韋","⾲":"韭","⾳":"音","⾴":"頁","⾵":"風","⾶":"飛","⾷":"食","⾸":"首","⾹":"香","⾺":"馬","⾻":"骨","⾼":"高","⾽":"髟","⾾":"鬥","⾿":"鬯","⿀":"鬲","⿁":"鬼","⿂":"魚","⿃":"鳥","⿄":"鹵","⿅":"鹿","⿆":"麥","⿇":"麻","⿈":"黃","⿉":"黍","⿊":"黒","⿋":"黹","⿌":"黽","⿍":"鼎","⿎":"鼓","⿏":"鼠","⿐":"鼻","⿑":"齊","⿒":"齒","⿓":"龍","⿔":"龜","⿕":"龠"}

関連

• メイリオの罠 - espresso3389の日記

FYIですが、UnicodeのNFKC正規化処理によって、康熙部首文字を漢字に戻すことができます。

Unicode 正規化すると「福神づけ」が「副神づけ」に、「①」が「1」なってしまうなど、余計な副作用が生じるのでよろしくないですよ。