(If you’re reading this via the OCaml planet, click the post title to see it with proper formatting – I shall fix this problem for the next post, I hope.)<\/p>\n
<\/p>\n
When calling our command line PDF Tools<\/a> to, for instance, merge files, one can use page specifiers with a simple grammar. For instance:<\/p>\n <\/p>\n cpdf file.pdf 1,2,6-end -o out.pdf<\/span><\/p>\n cpdf in.pdf 1,all -o fax.pdf<\/span><\/p>\n <\/p>\n (The second example is useful to duplicate the first page of a document as a fax cover sheet)<\/p>\n <\/p>\n If the grammar is simple, OCaml’s Genlex plus a small recursive parser is sufficient.<\/p>\n <\/p>\n Here’s an informal description of the mini language:<\/p>\n <\/p>\n \u2022 <\/span>A dash (-) de\ufb01nes ranges, e.g. <\/span>1-5 <\/span>or <\/span>6-3.\u00a0<\/span><\/p>\n \u2022 <\/span>A comma (,) allows one to specify several ranges, e.g. <\/span>1-2,4-5.\u00a0<\/span><\/p>\n \u2022 <\/span>The word\u00a0“<\/span>end” <\/span>represents the last page number.\u00a0<\/span><\/p>\n \u2022 <\/span>The words\u00a0“<\/span>odd” <\/span>and\u00a0“<\/span>even” <\/span>represent the odd and even pages.\u00a0<\/span><\/p>\n \u2022 <\/span>The word\u00a0“<\/span>reverse” <\/span>is the same as <\/span>end-1.\u00a0<\/span><\/p>\n \u2022 <\/span>The word\u00a0“<\/span>all”\u00a0<\/span>is the same as <\/span>1-end.\u00a0<\/span><\/p>\n \u2022 <\/span>A range must contain no spaces.\u00a0<\/span><\/p>\n Our input is the string in this language, our output an ordered list of page numbers. (I’m using some functions from the Utility module available with CamlPDF)<\/p>\n <\/p>\n First build a lexer:<\/p>\n <\/p>\n let lexer =<\/span><\/p>\n \u00a0 Genlex.make_lexer [“-“; “,”; “all”; “reverse”; “even”; “odd”; “end”]<\/span><\/p>\n <\/p>\n and define a utility function which lexes a string to a list of Genlex lexemes (no need to be lazy here):<\/p>\n <\/p>\n let lexwith s =<\/span><\/p>\n \u00a0 list_of_stream (lexer (Stream.of_string s))<\/span><\/p>\n <\/p>\n Now, this language is quite simple to deal with. We’ll pattern-match for all the simple cases – anything which doesn’t contain a comma is finite. If nothing matches, we assume the string contains one or more commas, and attempt to split it up.<\/p>\n <\/p>\n Here is the main function, which is given the end page of the PDF file to which the range applies (the start page is always 1), and a list of tokens to match against:<\/p>\n <\/p>\n let rec mk_numbers endpage = function<\/span><\/p>\n \u00a0 | [Genlex.Int n] -> [n]<\/span><\/p>\n \u00a0 | [Genlex.Int n; Genlex.Kwd “-“; Genlex.Int n’] -><\/span><\/p>\n \u00a0 \u00a0 \u00a0 if n > n’ then rev (ilist n’ n) else ilist n n’<\/span><\/p>\n \u00a0 | [Genlex.Kwd “end”; Genlex.Kwd “-“; Genlex.Int n] -><\/span><\/p>\n \u00a0 \u00a0 \u00a0 if n <= endpage<\/span><\/p>\n \u00a0 \u00a0 \u00a0 \u00a0 then rev (ilist n endpage)<\/span><\/p>\n \u00a0 \u00a0 \u00a0 \u00a0 else failwith “n > endpage”<\/span><\/p>\n \u00a0 | [Genlex.Int n; Genlex.Kwd “-“; Genlex.Kwd “end”] -><\/span><\/p>\n \u00a0 \u00a0 \u00a0 if n <= endpage<\/span><\/p>\n \u00a0 \u00a0 \u00a0 \u00a0 then ilist n endpage<\/span><\/p>\n \u00a0 \u00a0 \u00a0 \u00a0 else failwith “n > endpage2”<\/span><\/p>\n \u00a0 | [Genlex.Kwd “end”; Genlex.Kwd “-“; Genlex.Kwd “end”] -><\/span><\/p>\n \u00a0 \u00a0 \u00a0 [endpage]<\/span><\/p>\n \u00a0 | [Genlex.Kwd “even”] -><\/span><\/p>\n \u00a0\u00a0 \u00a0 \u00a0 drop_odds (ilist 1 endpage)<\/span><\/p>\n \u00a0 | [Genlex.Kwd “odd”] -><\/span><\/p>\n \u00a0\u00a0 \u00a0 \u00a0 really_drop_evens (ilist 1 endpage)<\/span><\/p>\n \u00a0 | [Genlex.Kwd “all”] -><\/span><\/p>\n \u00a0\u00a0 \u00a0 \u00a0 ilist 1 endpage<\/span><\/p>\n \u00a0 | [Genlex.Kwd “reverse”] -><\/span><\/p>\n \u00a0\u00a0 \u00a0 \u00a0 rev (ilist 1 endpage)<\/span><\/p>\n \u00a0 | toks -><\/span><\/p>\n \u00a0 \u00a0 \u00a0 let ranges = splitat_commas toks in<\/span><\/p>\n \u00a0\u00a0 \u00a0 \u00a0 \u00a0(* Check we’ve made progress *)<\/span><\/p>\n \u00a0\u00a0 \u00a0 \u00a0 \u00a0if ranges = [toks] then error “Bad page range” else\u00a0<\/span><\/p>\n \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 flatten (map (mk_numbers endpage) ranges)<\/span><\/p>\n <\/p>\n (ilist x y<\/span>\u00a0 produces the list [x, x + 1, … y – 1, y]<\/span>, \u00a0drop_odds [1;2;3;4;5;6;7]<\/span> is [2;4;6]<\/span>, really_drop_evens [1;2;3;4;5;6]<\/span> is [1;3;5]<\/span>)<\/p>\n <\/p>\n The auxilliary function to split a token list at commas into a list of token lists:<\/p>\n <\/p>\n let rec splitat_commas toks =<\/span><\/p>\n \u00a0 match cleavewhile (neq (Genlex.Kwd “,”)) toks with<\/span><\/p>\n \u00a0 | [], _ -> []<\/span><\/p>\n \u00a0\u00a0| some, [] -> [some]<\/span><\/p>\n \u00a0 | _::_ as before, _::rest -> before::splitat_commas rest\u00a0<\/span><\/p>\n <\/p>\n (cleavewhile<\/span> returns, in order, the elements at the beginning of a list until a predicate is false, paired with the rest of the elements, in order. \u00a0neq <\/span>is ( <> )<\/span>)<\/p>\n <\/p>\n And here’s the wrapper function, which unifies the error handling and tests for page numbers not within range.<\/p>\n <\/p>\n let parse_pagespec pdf spec =<\/span><\/p>\n \u00a0 let endpage = endpage_of_pdf pdf in<\/span><\/p>\n \u00a0 \u00a0 let numbers =<\/span><\/p>\n \u00a0 \u00a0 \u00a0 try mk_numbers endpage (lexwith spec) with<\/span><\/p>\n \u00a0\u00a0 \u00a0 \u00a0 \u00a0_ -> error (“Bad page specification ” ^ spec)<\/span><\/p>\n \u00a0 \u00a0 in<\/span><\/p>\n \u00a0 \u00a0 \u00a0 iter<\/span><\/p>\n \u00a0 \u00a0 \u00a0 \u00a0 (fun n -> if n <> endpage then<\/span><\/p>\n \u00a0\u00a0 \u00a0 \u00a0 \u00a0 \u00a0 error (“Page ” ^ string_of_int n ^ ” does not exist.”))<\/span><\/p>\n \u00a0 \u00a0 \u00a0 \u00a0 numbers;<\/span><\/p>\n \u00a0 \u00a0 \u00a0 numbers<\/span><\/p>\n This is often easier to write and debug than using a separate tool intended for highly complex grammars.<\/span><\/p>\n <\/span><\/div>\n","protected":false},"excerpt":{"rendered":" (If you’re reading this via the OCaml planet, click the post title to see it with proper formatting – I shall fix this problem for the next post, I hope.) When calling our command line PDF Tools to, for instance, … Continue reading
<\/span><\/p>\n
<\/span><\/p>\n
<\/span><\/p>\n