Archives
/
gnosygnu_xowa
mirror of https://github.com/gnosygnu/xowa
1
0
Fork 0
Code Issues Projects Releases Wiki Activity

Mw_parse: Add equivalent of htmlspecialchar

Browse Source
pull/620/head
gnosygnu 8 years ago
parent 6ee274efd6
commit 25f74d7d80
5 changed files with 81 additions and 37 deletions
Show Stats Download Patch File Download Diff File

34
100_core/src/gplx/Bry_bfr.java
Unescape View File

@ -17,6 +17,7 @@ along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
package gplx;
import gplx.core.primitives.*; import gplx.core.brys.*; import gplx.core.encoders.*;
import gplx.langs.htmls.entitys.*;
public class Bry_bfr {
private Bry_bfr_mkr_mgr mkr_mgr; private int reset;
public byte[] Bfr() {return bfr;} private byte[] bfr;
@ -293,7 +294,38 @@ public class Bry_bfr {
}
}
if (clean)
Add(val);
Add_mid(val, bgn, end);
return this;
}
public Bry_bfr Add_bry_escape_html(byte[] val) {return Add_bry_escape_html(val, 0, val.length);}
public Bry_bfr Add_bry_escape_html(byte[] val, int bgn, int end) { // uses PHP rules for htmlspecialchars; REF.PHP:http://php.net/manual/en/function.htmlspecialchars.php
boolean clean = true;
for (int i = bgn; i < end; ++i) {
byte[] escaped = null;
byte b = val[i];
switch (b) {
case Byte_ascii.Amp: escaped = Gfh_entity_.Amp_bry; break;
case Byte_ascii.Quote: escaped = Gfh_entity_.Quote_bry; break;
case Byte_ascii.Apos: escaped = Gfh_entity_.Apos_num_bry; break;
case Byte_ascii.Lt: escaped = Gfh_entity_.Lt_bry; break;
case Byte_ascii.Gt: escaped = Gfh_entity_.Gt_bry; break;
}
if (escaped == null && clean) {
continue;
}
else {
if (clean) {
clean = false;
this.Add_mid(val, bgn, i);
}
if (escaped == null)
this.Add_byte(b);
else
this.Add(escaped);
}
}
if (clean)
Add_mid(val, bgn, end);
return this;
}
public Bry_bfr Add_str_u8_w_nl(String s) {Add_str_u8(s); return Add_byte_nl();}

27
100_core/src/gplx/Bry_bfr_tst.java
Unescape View File

@ -16,7 +16,7 @@ You should have received a copy of the GNU Affero General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
package gplx;
import org.junit.*;
import org.junit.*; import gplx.core.tests.*;
public class Bry_bfr_tst {
private Bry_bfr bb = Bry_bfr_.New();
@Before public void setup() {bb.Clear();} private ByteAryBfr_fxt fxt = new ByteAryBfr_fxt();
@ -187,9 +187,16 @@ public class Bry_bfr_tst {
fxt.Test_Add_int_pad_bgn(Byte_ascii.Num_0, 3, 1000, "1000");
}
@Test public void Add_bry_escape() {
fxt.Test_Add_bry_escape("abc" , "abc"); // nothing to escape
fxt.Test_Add_bry_escape("a'bc" , "a''bc"); // single escape (code handles first quote differently)
fxt.Test_Add_bry_escape("a'b'c" , "a''b''c"); // double escape (code handles subsequent quotes different than first)
fxt.Test__add_bry_escape("abc" , "abc"); // nothing to escape
fxt.Test__add_bry_escape("a'bc" , "a''bc"); // single escape (code handles first quote differently)
fxt.Test__add_bry_escape("a'b'c" , "a''b''c"); // double escape (code handles subsequent quotes different than first)
fxt.Test__add_bry_escape("abc", 1, 2 , "b"); // nothing to escape
}
@Test public void Add_bry_escape_html() {
fxt.Test__add_bry_escape_html("abc" , "abc"); // escape=none
fxt.Test__add_bry_escape_html("a&\"'<>b" , "a&amp;&quot;&#39;&lt;&gt;b"); // escape=all; code handles first escape differently
fxt.Test__add_bry_escape_html("a&b&c" , "a&amp;b&amp;c"); // staggered; code handles subsequent escapes differently
fxt.Test__add_bry_escape_html("abc", 1, 2 , "b"); // by index; fixes bug in initial implementation
}
@Test public void Insert_at() {
fxt.Test_Insert_at("abcd", 0, "xyz" , "xyzabcd"); // bgn
@ -217,13 +224,15 @@ public class Bry_bfr_tst {
}
class ByteAryBfr_fxt {
private final Bry_bfr bfr = Bry_bfr_.Reset(16);
public Bry_bfr Bfr() {return bfr;}
public void Clear() {
bfr.ClearAndReset();
}
public void Test_Add_int_pad_bgn(byte pad_byte, int str_len, int val, String expd) {Tfds.Eq(expd, bfr.Add_int_pad_bgn(pad_byte, str_len, val).To_str_and_clear());}
public void Test_Add_bry_escape(String val, String expd) {
byte[] val_bry = Bry_.new_u8(val);
Tfds.Eq(expd, bfr.Add_bry_escape(Byte_ascii.Apos, Byte_.Ary(Byte_ascii.Apos, Byte_ascii.Apos), val_bry, 0, val_bry.length).To_str_and_clear());
public void Test__add_bry_escape(String src, String expd) {Test__add_bry_escape(src, 0, String_.Len(src), expd);}
public void Test__add_bry_escape(String src, int src_bgn, int src_end, String expd) {
byte[] val_bry = Bry_.new_u8(src);
Tfds.Eq(expd, bfr.Add_bry_escape(Byte_ascii.Apos, Byte_.Ary(Byte_ascii.Apos, Byte_ascii.Apos), val_bry, src_bgn, src_end).To_str_and_clear());
}
public void Test_Insert_at(String init, int pos, String val, String expd) {Tfds.Eq(expd, bfr.Add_str_u8(init).Insert_at(pos, Bry_.new_u8(val)).To_str_and_clear());}
public void Test_Insert_at(String init, int pos, String val, int val_bgn, int val_end, String expd) {Tfds.Eq(expd, bfr.Add_str_u8(init).Insert_at(pos, Bry_.new_u8(val), val_bgn, val_end).To_str_and_clear());}
@ -233,4 +242,8 @@ class ByteAryBfr_fxt {
public void Test__to_bry_ary_and_clear(String bfr_str, String... expd) {
Tfds.Eq_ary(expd, String_.Ary(bfr.Add_str_u8(bfr_str).To_bry_ary_and_clear()));
}
public void Test__add_bry_escape_html(String src, String expd) {Test__add_bry_escape_html(src, 0, String_.Len(src), expd);}
public void Test__add_bry_escape_html(String src, int src_bgn, int src_end, String expd) {
Gftest.Eq__bry(Bry_.new_u8(expd), bfr.Add_bry_escape_html(Bry_.new_u8(src), src_bgn, src_end).To_bry_and_clear());
}
}

2
400_xowa/src/gplx/xowa/addons/bldrs/mass_parses/parses/wkrs/Xomp_parse_wkr.java
Unescape View File

@ -111,7 +111,7 @@ public class Xomp_parse_wkr implements Gfo_invk {
// if ns changed and prv_ns is main
if (cur_ns != prv_ns) {
if (prv_ns == gplx.xowa.wikis.nss.Xow_ns_.Tid__main)
wiki.Cache_mgr().Free_mem__all(); // NOTE: clears page and wbase cache only; needed else OutOfMemory error for en.w in 25th hour; DATE:2017-01-07
wiki.Cache_mgr().Free_mem__all(); // NOTE: clears page and wbase cache only; needed else OutOfMemory error for en.w in 25th hour; DATE:2017-01-11
prv_ns = cur_ns;
}
Xoae_page wpg = Xoae_page.New(wiki, ttl);

49
400_xowa/src/gplx/xowa/parsers/mws/prepros/Xomw_prepro_wkr.java
Unescape View File

@ -132,11 +132,11 @@ public class Xomw_prepro_wkr { // TS.UNSAFE:caching for repeated calls
int start_pos = Bry_find_.Find_fwd(src, Bry__only_include_bgn, i, src_len);
if (start_pos == Bry_find_.Not_found) {
// Ignored section runs to the end
accum.Add_str_a7("<ignore>").Add(htmlspecialchars(Bry_.Mid(src, i))).Add_str_a7("</ignore>");
accum.Add_str_a7("<ignore>").Add_bry_escape_html(src, i, src_len).Add_str_a7("</ignore>");
break;
}
int tag_end_pos = start_pos + Bry__only_include_bgn.length; // past-the-end
accum.Add_str_a7("<ignore>").Add(htmlspecialchars(Bry_.Mid(src, i, tag_end_pos))).Add_str_a7("</ignore>");
accum.Add_str_a7("<ignore>").Add_bry_escape_html(src, i, tag_end_pos).Add_str_a7("</ignore>");
i = tag_end_pos;
find_only_include = false;
}
@ -205,7 +205,7 @@ public class Xomw_prepro_wkr { // TS.UNSAFE:caching for repeated calls
literal_len++;
}
if (literal_len > 0) {
accum.Add(htmlspecialchars(Bry_.Mid(src, i, i + literal_len)));
accum.Add_bry_escape_html(src, i, i + literal_len);
i += literal_len;
}
if (i >= src_len) {
@ -277,7 +277,7 @@ public class Xomw_prepro_wkr { // TS.UNSAFE:caching for repeated calls
int end_pos = Bry_find_.Find_fwd(src, Bry__comment_end, i + 4, src_len);
if (end_pos == Bry_find_.Not_found) {
// Unclosed comment in input, runs to end
accum.Add_str_a7("<comment>").Add(htmlspecialchars(Bry_.Mid(src, i))).Add_str_a7("</comment>");
accum.Add_str_a7("<comment>").Add_bry_escape_html(src, i, src_len).Add_str_a7("</comment>");
i = src_len;
}
else {
@ -331,7 +331,7 @@ public class Xomw_prepro_wkr { // TS.UNSAFE:caching for repeated calls
break;
}
inner = Bry_.Mid(src, bgn_pos, end_pos);
accum.Add_str_a7("<comment>").Add(htmlspecialchars(inner)).Add_str_a7("</comment>");
accum.Add_str_a7("<comment>").Add_bry_escape_html(inner).Add_str_a7("</comment>");
}
// Do a line-start run next time to look for headings after the comment
@ -353,7 +353,7 @@ public class Xomw_prepro_wkr { // TS.UNSAFE:caching for repeated calls
}
i = end_pos + 1;
inner = Bry_.Mid(src, bgn_pos, end_pos + 1);
accum.Add_str_a7("<comment>").Add(htmlspecialchars(inner)).Add_str_a7("</comment>");
accum.Add_str_a7("<comment>").Add_bry_escape_html(inner).Add_str_a7("</comment>");
continue;
}
}
@ -375,7 +375,7 @@ public class Xomw_prepro_wkr { // TS.UNSAFE:caching for repeated calls
// Handle ignored tags
if (ignored_tags.Has(name)) {
accum.Add_str_a7("<ignore>").Add(htmlspecialchars(Bry_.Mid(src, i, tag_end_pos + 1))).Add_str_a7("</ignore>");
accum.Add_str_a7("<ignore>").Add_bry_escape_html(src, i, tag_end_pos + 1).Add_str_a7("</ignore>");
i = tag_end_pos + 1;
continue;
}
@ -426,7 +426,7 @@ public class Xomw_prepro_wkr { // TS.UNSAFE:caching for repeated calls
&& elem_end_found) {
inner = Bry_.Mid(src, tag_end_pos + 1, elem_end_lhs);
i = elem_end_rhs;
tmp_bfr.Add_str_a7("<close>").Add(htmlspecialchars(Bry_.Mid(src, elem_end_lhs, elem_end_rhs))).Add_str_a7("</close>");
tmp_bfr.Add_str_a7("<close>").Add_bry_escape_html(src, elem_end_lhs, elem_end_rhs).Add_str_a7("</close>");
close = tmp_bfr.To_bry_and_clear();
}
else {
@ -440,7 +440,7 @@ public class Xomw_prepro_wkr { // TS.UNSAFE:caching for repeated calls
else {
// Don't match the tag, treat opening tag as literal and resume parsing.
i = tag_end_pos + 1;
accum.Add(htmlspecialchars(Bry_.Mid(src, tag_bgn_pos, tag_end_pos + 1)));
accum.Add_bry_escape_html(src, tag_bgn_pos, tag_end_pos + 1);
// Cache results, otherwise we have O(N^2) performance for input like <foo><foo><foo>...
no_more_closing_tag.Add_if_dupe_use_nth(name, name);
continue;
@ -450,18 +450,26 @@ public class Xomw_prepro_wkr { // TS.UNSAFE:caching for repeated calls
// <includeonly> and <noinclude> just become <ignore> tags
if (ignored_elements.Has(name)) {
accum.Add_str_a7("<ignore>").Add(htmlspecialchars(Bry_.Mid(src, tag_bgn_pos, i))).Add_str_a7("</ignore>");
accum.Add_str_a7("<ignore>").Add_bry_escape_html(src, tag_bgn_pos, i).Add_str_a7("</ignore>");
continue;
}
accum.Add_str_a7("<ext>");
byte[] atr_bry = atr_end <= atr_bgn ? Bry_.Empty : Bry_.Mid(src, atr_bgn, atr_end);
// PORTED:
// if ( $attrEnd <= $attrStart ) {
// $attr = '';
// } else {
// $attr = substr( $text, $attrStart, $attrEnd - $attrStart );
// }
accum.Add_str_a7("<name>").Add(name).Add_str_a7("</name>");
// Note that the attr element contains the whitespace between name and attribute,
// this is necessary for precise reconstruction during pre-save transform.
accum.Add_str_a7("<attr>").Add(htmlspecialchars(atr_bry)).Add_str_a7("</attr>");
accum.Add_str_a7("<attr>");
if (atr_end > atr_bgn)
accum.Add_bry_escape_html(src, atr_bgn, atr_end);
accum.Add_str_a7("</attr>");
if (inner != null) {
accum.Add_str_a7("<inner>").Add(htmlspecialchars(inner)).Add_str_a7("</inner>");
accum.Add_str_a7("<inner>").Add_bry_escape_html(inner).Add_str_a7("</inner>");
}
accum.Add(close).Add_str_a7("</ext>");
}
@ -582,7 +590,8 @@ public class Xomw_prepro_wkr { // TS.UNSAFE:caching for repeated calls
}
else {
// Add literal brace(s)
accum.Add(htmlspecialchars(Bry_.Repeat_bry(cur_char, count)));
for (int j = 0; j < count; j++)
accum.Add_bry_escape_html(cur_char);
}
i += count;
}
@ -613,7 +622,8 @@ public class Xomw_prepro_wkr { // TS.UNSAFE:caching for repeated calls
if (matching_count <= 0) {
// No matching element found in callback array
// Output a literal closing brace and continue
accum.Add(htmlspecialchars(Bry_.Repeat_bry(cur_char, count)));
for (int j = 0; j < count; j++)
accum.Add_bry_escape_html(cur_char);
i += count;
continue;
}
@ -718,15 +728,6 @@ public class Xomw_prepro_wkr { // TS.UNSAFE:caching for repeated calls
root_accum.Add_str_a7("</root>");
return root_accum.To_bry_and_clear();
}
private byte[] htmlspecialchars(byte[] bry) {
// http://php.net/manual/en/function.htmlspecialchars.php
//& (ampersand) &amp;
//" (double quote) &quot;, unless ENT_NOQUOTES is set
//' (single quote) &#039; (for ENT_HTML401) or &apos; (for ENT_XML1, ENT_XHTML or ENT_HTML5), but only when ENT_QUOTES is set
//< (less than) &lt;
//> (greater than) &gt;
return bry;
}
private Xomw_prepro_rule Get_rule(byte[] bry) {
if (Bry_.Eq(bry, rule_curly.bgn)) return rule_curly;
else if (Bry_.Eq(bry, rule_brack.bgn)) return rule_brack;

6
400_xowa/src/gplx/xowa/parsers/mws/prepros/Xomw_prepro_wkr__tst.java
Unescape View File

@ -32,15 +32,13 @@ public class Xomw_prepro_wkr__tst {
fxt.Test__parse("a{{{b}}}c", "<root>a<tplarg lineStart=\"1\"><title>b</title></tplarg>c</root>");
}
@Test public void Comment() {
fxt.Test__parse("a<!--b-->c", "<root>a<comment><!--b--></comment>c</root>");
fxt.Test__parse("a<!--b-->c", "<root>a<comment>&lt;!--b--&gt;</comment>c</root>");
}
@Test public void Ext__pre() {
fxt.Test__parse("a<pre id=\"1\">b</pre>c", "<root>a<ext><name>pre</name><attr> id=\"1\"</attr><inner>b</inner><close></pre></close></ext>c</root>");
fxt.Test__parse("a<pre id=\"1\">b</pre>c", "<root>a<ext><name>pre</name><attr> id=&quot;1&quot;</attr><inner>b</inner><close>&lt;/pre&gt;</close></ext>c</root>");
}
/*
TODO:
* htmlspecialchars
* for_inclusion; <onlyinclude> in String
* heading.general
* heading.EOS: "==a" (no closing ==)

Write Preview
Loading…
Cancel
Save