ref: 8a46d62744d2969ba9c3488672027509f00fdb4b
parent: c65c9894c643308c87de69de8c0be5c6c6e1c663
author: Ori Bernstein <[email protected]>
date: Wed Oct 23 08:51:09 EDT 2013
Add beginning/end of line matches
--- a/compile.myr
+++ b/compile.myr
@@ -24,8 +24,8 @@
/* meta */
`Cap tree#
- `Astart
- `Aend
+ `Bol /* beginning of line */
+ `Eol /* end of line */
;;
type parseresult = union
@@ -76,6 +76,12 @@
`Dot: append(re, `Idot);;
/* meta */
+ `Bol:
+ append(re, `Ibol)
+ ;;
+ `Eol:
+ append(re, `Ibol)
+ ;;
`Cap a:
m = re.nmatch++
append(re, `Ilbra m)
@@ -164,32 +170,19 @@
std.put("%i:\t", i)match re.prog[i]
/* Char matching. Consume exactly one byte from the string. */
- `Ibyte b:
- std.put("`Ibyte %b (%c)\n", b, b castto(char))- ;;
- `Irange (start, end):
- std.put("`Irange (%b,%b)\n", start, end)- ;;
- `Idot:
- std.put("`Idot\n")- ;;
+ `Ibyte b: std.put("`Ibyte %b (%c)\n", b, b castto(char)) ;;+ `Irange (start, end): std.put("`Irange (%b,%b)\n", start, end) ;;+ `Idot: std.put("`Idot\n") ;;/* capture groups */
- `Ilbra m:
- std.put("`Ilbra %z\n", m)- ;;
- `Irbra m:
- std.put("`Irbra %z\n", m)- ;;
+ `Ilbra m: std.put("`Ilbra %z\n", m) ;;+ `Irbra m: std.put("`Irbra %z\n", m) ;;+ /* anchors */
+ `Ibol: std.put("`Ibol\n");;+ `Ieol: std.put("`Ieol\n");;/* control flow */
- `Ifork (lip, rip):
- std.put("`Ifork (%z,%z)\n", lip, rip)- ;;
- `Ijmp ip:
- std.put("`Ijmp %z\n", ip)- ;;
- `Imatch:
- std.put("`Imatch\n")- ;;
+ `Ifork (lip, rip): std.put("`Ifork (%z,%z)\n", lip, rip) ;;+ `Ijmp ip: std.put("`Ijmp %z\n", ip) ;;+ `Imatch: std.put("`Imatch\n") ;;;;
;;
}
@@ -224,6 +217,12 @@
std.put("Quest\n")dump(a, indent + 1)
;;
+ `Bol:
+ std.put("Bol\n")+ ;;
+ `Eol:
+ std.put("Eol\n")+ ;;
/* end matches */
`Class sl:
std.put("Class [a..b]\n")@@ -342,8 +341,8 @@
'?': -> `Fail (`Badrep);;
'[': ret = chrclass(re);;
'.': getc(re); ret = mk(`Dot);;
- '^': getc(re); ret = mk(`Astart);;
- '$': getc(re); ret = mk(`Aend);;
+ '^': getc(re); ret = mk(`Bol);;
+ '$': getc(re); ret = mk(`Eol);;
'(': getc(re)
match altexpr(re)
--- a/interp.myr
+++ b/interp.myr
@@ -92,6 +92,23 @@
Non-consuming. All of these recursively call step() until
exactly one byte is consumed from the string.
*/
+ `Ibol:
+ trace(re, "\t%i:\tBol\n", thr.ip)
+ if re.strp == 0 || str[re.strp -1] == 0x10
+ thr.ip++
+ step(re, tid)
+ else
+ kill(re, tid, "not beginning of line")
+ ;;
+ ;;
+ `Ieol:
+ trace(re, "\t%i:\tEol\n", thr.ip)
+ if re.strp == str.len || str[re.strp] == 0x10
+ step(re, tid)
+ else
+ kill(re, tid, "not end of line")
+ ;;
+ ;;
`Ilbra m:
trace(re, "\t%i:\tLbra %z\n", thr.ip, m)
trace(re, "\t\tmatch start = %z\n", re.strp)
--- a/main.myr
+++ b/main.myr
@@ -3,9 +3,9 @@
const main = {var found
- match regex.compile("(a*b)*")+ match regex.compile(".*^a*")`std.Success re:
- found = regex.exec(re, "a")
+ found = regex.exec(re, "b\naaa")
std.put("Found = %t: len = %z\n", found, re.strp)-> 0
;;
--- a/types.myr
+++ b/types.myr
@@ -42,6 +42,10 @@
`Ilbra std.size
`Irbra std.size
+ /* anchors */
+ `Ibol
+ `Ieol
+
/* control flow */
`Ifork [std.size, std.size]
`Ijmp std.size