add new c1x standard draft (n1570), +pdf2html.sh script

[c-standard] / translit.sh

diff --git a/translit.sh b/translit.sh
index 0bf85a0..d880571 100755 (executable)
--- a/translit.sh
+++ b/translit.sh
@@ -30,7 +30,7 @@ s/≤/<=/g
 s/≥/>=/g
 s/≠/!=/g
 s/Σ/(Sum)/g
-s/√/sqrt:/g
+s/√/(sqrt)/g
 s/π/pi/g
 s/∞/(inf)/g
 s/ƒ/fl./g
@@ -41,19 +41,35 @@ s/•/o/g
 s/⎯/-/g
 s/↑/(uparrow)/g
 s/↓/(downarrow)/g
-s/↔/(<->)/g
-s/→/(->)/g
+s/↔/<->/g
+s/→/->/g
 s/‘/'\''/g
 s/’/'\''/g
 # pdftotext layout fixes
 s/_ _/__/g
-s/\([0-9]\). \([0-9]\)/\1.\2/g
+# floats are sometimes broken
+s/\([0-9]\)\. \([0-9]\)/\1.\2/g
 ' | LC_ALL=C tr -c '\n-~' '?' | awk '
 BEGIN {
        getline
        last=$0
        side=0
 }
+/^$/ {
+       nl=nl "\n"
+       next
+}
+# TODO: shift page numbers
+#function inc(x) {
+#      if (x ~ /[0-9]/)
+#              return x+1
+#      if (sub(/viii$/,"ix",x) ||
+#          sub(/iii$/,"iv",x) ||
+#          sub(/iv$/,"v",x) ||
+#          sub(/ix$/,"x",x))
+#              return x
+#      return x "i"
+#}
 /^\(newpage\)/ {
        n=split(last,a)
        if(side)
@@ -61,7 +77,10 @@ BEGIN {
        else
                p=a[n]
        side=!side
-       print "[page " p "]"
+#      if (p !~ /[0-9]/ && $0 ~ /INTERNATIONAL STANDARD/)
+#              p=0
+#      print "\n[page " inc(p) "]"
+       print "\n[page " p "]"
        getline
        getline
        last=$0
@@ -69,6 +88,6 @@ BEGIN {
 }
 {
        print last
-       last=$0
-}
-'
+       last=nl $0
+       nl=""
+}'