(* USE WITH CAUTION BECAUSE OLD CAPTIONS WILL BE OVERWRITTEN! *)

(***** Change these lines *****)
minyear=2000;
maxyear=2000;
minroll=114;
maxroll=114;
dir="e:\\public_html\\pictures\\archives";
(******************************)

BeginPackage["takebetween`"]

takebetween::usage =
   "If the string s mathes the pattern *s1*s2*
then takebetween[s,s1,s2][[1]] is True
and takebetween[s,s1,s2][[2]] is the string between s1 and s2,
otherwise takebetween[s,s1,s2] is {False,{}}."

Begin["`private`"]

takebetween[s_String,s1_String,s2_String]:=
If[StringMatchQ[s,"*"<>s1<>"*"<>s2<>"*",IgnoreCase->True],
posit=StringPosition[s,s1,IgnoreCase->True];
drop1={1,posit[[1,2]]};
s12=StringDrop[s,drop1];
posit=StringPosition[s12,s2,IgnoreCase->True];
drop2={posit[[1,1]],StringLength[s12]};
s12=StringDrop[s12,drop2];
{True,s12},{False,{}}]

End[]

EndPackage[]


Do[ (* year *) year=ToString[nyear];
Do[ (* roll *) roll=ToString[nroll];
dirroll=ToFileName[{dir,year,roll}];
If[SetDirectory[dirroll]!=StringDrop[dirroll,-1],Goto[1]];
browser0=ToFileName[dirroll,"browser0.htm"];
capt=ToFileName[dirroll,"captions.txt"];
Print["   ----------   ",capt];
data=ReadList[browser0,String];
mlines=Length[data];
oc=OpenWrite[capt,PageWidth->Infinity];
captnumb="";
Do[line=data[[nline]];
t=takebetween[line,"<TITLE>","</TITLE>"];
uline=ToUpperCase[line];
If[ t[[1]],
title=StringReplace[t[[2]],": thumbnail preview of photo shots"->""];
Write[oc,"#ROLL"//OutputForm];
Write[oc,title//OutputForm];
Print["#ROLL\n",title];
Goto[2];
];
line0=StringReplace[uline," "->""];
If[ StringMatchQ[line0,"*<TD>@&NBSP;</TD>"],
captnumb=takebetween[line0,"<TD>","&NBSP;</TD>"][[2]];c="";Goto[2] ];
If[captnumb!="",c=c<>"\n"<>line];
t=takebetween[c,"<TD>&nbsp;","</TD>"];
If[ t[[1]],
c=t[[2]];
c="#"<>captnumb<>": "<>c;
captnumb="";
Write[oc,c//OutputForm];
Print[c];
];
Label[2],{nline,mlines}];
Close[oc];
Label[1],{nroll,minroll,maxroll}],{nyear,minyear,maxyear}];