llpp/parser.ml

327 lines
8.9 KiB
OCaml

(* based on Tor Andersson's XML parser from MuPDF's XPS module *)
let iswhite = function
| '\r' | '\n' | '\t' | ' ' -> true
| _ -> false
let isname = function
| '.' | '-' | '_' | ':' -> true
| c -> (c >= '0' && c <= '9')
|| (c >= 'a' && c <= 'z')
|| (c >= 'A' && c <= 'Z')
exception Parse_error of string * string * int
let parse_error msg s pos = raise (Parse_error (msg, s, pos))
let enent s pos len =
let b = Buffer.create len in
let rec loop i =
if i - pos = len
then Buffer.contents b
else (
begin match s.[i] with
| '<' -> Buffer.add_string b "&lt;"
| '>' -> Buffer.add_string b "&gt;"
| '\'' -> Buffer.add_string b "&apos;"
| '\"' -> Buffer.add_string b "&quot;"
| '&' -> Buffer.add_string b "&amp;"
| c ->
let code = Char.code c in
if code < 32 || code > 127
then (
Buffer.add_string b "&#";
Buffer.add_string b (string_of_int code);
Buffer.add_char b ';';
)
else Buffer.add_char b c
end;
loop (i+1)
)
in
loop pos
let unent b s pos len =
let rec loop i =
if i != pos + len
then
let amppos =
try
String.index_from s i '&'
with Not_found -> -1
in
if amppos = -1 || amppos >= pos + len
then Buffer.add_substring b s i (pos + len - i)
else (
Buffer.add_substring b s i (amppos - i);
if amppos = i + len then Utils.error "lonely amp";
let semipos =
try
let semipos = String.index_from s (amppos+1) ';' in
if semipos >= pos + len then raise Not_found;
semipos
with Not_found ->
Utils.error "amp not followed by semicolon at %d" amppos
in
let subslen = semipos-amppos-1 in
if subslen = 0 then Utils.error "empty amp at %d" amppos;
let subs = String.sub s (amppos+1) subslen in
if subs.[0] = '#'
then (
if subslen = 1
then Utils.error "empty amp followed by hash at %d" amppos;
let code =
if subs.[1] = 'x'
then Scanf.sscanf subs "#x%x" (fun n -> n)
else int_of_string (String.sub subs 1 (subslen-1))
in
let c = Char.unsafe_chr code in
Buffer.add_char b c
)
else (
match subs with
| "lt" -> Buffer.add_char b '<'
| "gt" -> Buffer.add_char b '>'
| "amp" -> Buffer.add_char b '&'
| "apos" -> Buffer.add_char b '\''
| "quot" -> Buffer.add_char b '\"'
| _ -> Utils.error "unknown amp %S" subs
);
loop (semipos+1)
)
in
loop pos
let subs s pos =
let len = String.length s in
let left = len - pos in
if left < 0
then Printf.sprintf "(pos=%d len=%d left=%d)" pos len left
else
let len = min left 10 in
String.sub s pos len
let ts = function
| `text -> "text"
| `lt -> "lt"
| `close -> "close"
| `exclam -> "exclam"
| `question -> "question"
| `doctype -> "doctype"
| `comment -> "comment"
| `tag -> "tag"
type attr = string * string
and attrs = attr list
and vp =
| Vdata
| Vcdata
| Vopen of string * attrs * bool
| Vclose of string
| Vend
and 'a v = { f : 'a v -> vp -> int -> int -> 'a v; accu : 'a }
let parse v s =
let r_comment_terminator = Str.regexp "-->"
and r_CDATA_terminator = Str.regexp "\\]\\]>"
and r_q_terminator = Str.regexp "\\?>" in
let slen = String.length s in
let find_substr pos subs r =
let pos =
try
Str.search_forward r s pos
with Not_found ->
parse_error ("cannot find substring " ^ subs) s pos
in
pos
in
let begins_with pos prefix = Utils.substratis s pos prefix in
let find_non_white pos =
let rec forward i =
if i >= slen
then parse_error "cannot find non white space character" s pos;
if iswhite s.[i] then forward (i+1) else i in
forward pos
in
let getname pos =
let non_name_pos =
let rec find_non_name i =
if i >= slen then parse_error "cannot find non name character" s pos;
if isname s.[i] then find_non_name (i+1) else i
in
find_non_name pos
in
non_name_pos, String.sub s pos (non_name_pos - pos)
in
let rec collect v pos t =
if pos >= slen && t != `text
then parse_error ("not enough data for " ^ ts t) s pos;
match t with
| `text ->
let ltpos =
try
String.index_from s pos '<'
with Not_found ->
let rec trailsbywhite i =
if pos+i = String.length s
then -1
else (
if not (iswhite s.[pos+i])
then parse_error "garbage at the end" s pos
else trailsbywhite (i+1)
)
in
trailsbywhite 0
in
if ltpos = -1
then v.f v Vend pos slen, slen
else
let start_of_text_pos = find_non_white pos in
let end_of_text_pos =
if start_of_text_pos < ltpos
then
let rec find i =
if i = start_of_text_pos || not (iswhite s.[i])
then i+1
else find (i-1)
in
find (ltpos-1)
else start_of_text_pos
in
let v =
if start_of_text_pos != end_of_text_pos
then v.f v Vdata start_of_text_pos end_of_text_pos
else v
in
collect v (ltpos+1) `lt
| `lt ->
let pos, t =
match s.[pos] with
| '/' -> (pos+1), `close
| '!' -> (pos+1), `exclam
| '?' -> (pos+1), `question
| c when isname c -> pos, `tag
| _ -> parse_error "invalid data after <" s pos
in
collect v pos t
| `close ->
let tag_name_pos = find_non_white pos in
let tag_name_end_pos, close_tag_name = getname tag_name_pos in
let close_tag_pos = find_non_white tag_name_end_pos in
if s.[close_tag_pos] != '>'
then parse_error "missing >" s pos;
let pos' = close_tag_pos + 1 in
let v = v.f v (Vclose close_tag_name) pos pos' in
collect v pos' `text
| `doctype ->
let close_tag_pos =
try
String.index_from s pos '>'
with Not_found ->
parse_error "doctype is not terminated" s pos
in
collect v (close_tag_pos+1) `text
| `comment ->
let pos =
try
find_substr pos "-->" r_comment_terminator
with Not_found ->
parse_error "comment is not terminated" s pos
in
collect v (pos+3) `text
| `exclam ->
if begins_with pos "[CDATA["
then
let cdata_start = pos+7 in
let cdata_end = find_substr cdata_start "]]>" r_CDATA_terminator in
let v = v.f v Vcdata cdata_start cdata_end in
collect v (cdata_end+3) `text
else (
if begins_with pos "DOCTYPE"
then
collect v (pos+7) `doctype
else (
if begins_with pos "--"
then collect v (pos+2) `comment
else parse_error "unknown shit after exclamation mark" s pos
)
)
| `question ->
let pos = find_substr pos "?>" r_q_terminator in
collect v (pos+2) `text
| `tag ->
let pos', name = getname pos in
let attrs, pos', closed = collect_attributes pos' in
let v = v.f v (Vopen (name, attrs, closed)) pos pos' in
collect v pos' `text
and collect_attributes pos =
let rec f accu pos =
let nameval pos =
let pos, name = getname pos in
let pos = find_non_white pos in
if s.[pos] = '='
then
let qpos = pos+1 in
if qpos = slen
then parse_error "not enough data for attribute" s pos;
let qc = s.[qpos] in
if not (qc = '\'' || qc = '\"')
then parse_error "assignment is not followed by a quote" s pos;
let closing_q_pos =
let rec find i =
if i = slen
then parse_error "not enough data for attribute value" s pos;
if s.[i] = qc then i else find (i+1)
in
find (qpos+1)
in
let vallen = closing_q_pos - (qpos+1) in
let val' = String.sub s (qpos+1) vallen in
(name, val'), closing_q_pos+1
else parse_error "attribute name not followed by '='" s pos
in
let pos = find_non_white pos in
if s.[pos] = '>'
then accu, pos+1, false
else (
if slen - pos > 2 && s.[pos] = '/' && s.[pos+1] = '>'
then accu, pos+2, true
else (
if isname s.[pos]
then (
let nameval, pos = nameval pos in
let accu = nameval :: accu in
f accu pos
)
else parse_error "malformed attribute list" s pos;
)
)
in
f [] pos
in
let _, _ = collect v 0 `text in
v.accu