-//\r
-// This software is now distributed according to\r
-// the Lesser Gnu Public License. Please see\r
-// http://www.gnu.org/copyleft/lesser.txt for\r
-// the details.\r
-// -- Happy Computing!\r
-//\r
-package com.stevesoft.pat;\r
-\r
-import com.stevesoft.pat.wrap.*;\r
-\r
-/** Like Skip, but implements a\r
- <a href="http://www.dcc.uchile.cl/~rbaeza/handbook/algs/7/713b.srch.p.html">\r
- Boyer-Moore-Horspool</a> type search\r
- method that has been modified to be more like a "T-search" (see\r
- the Michael Tamm''s article in <i>C'T, magazin fuer computer und technic</i>, August 97\r
- p 292). Yet another important source of information for me was\r
- the <a href="http://www.go2net.com/people/paulp/deep/1997/05/14/">\r
- Deep Magic</a> article on string searching. As of this writing, I can\r
- beat String's indexOf method in many cases.\r
- @see com.stevesoft.pat.Skip\r
- @see com.stevesoft.pat.Skip2\r
- */\r
-public class SkipBMH\r
- extends Skip\r
-{\r
- // This number could be 256, but I think it's\r
- // big enough. Note, it must be a power of 2.\r
- final int MAX_CHAR = 64;\r
- final char[] skip = new char[MAX_CHAR];\r
- int sm1;\r
- int jump_ahead = 0;\r
- char uc, lc, tc, x;\r
- final boolean exact(char c)\r
- {\r
- return (ign && anyc(c)) || c == x;\r
- }\r
-\r
- final boolean anyc(char c)\r
- {\r
- return c == uc || c == lc || c == tc;\r
- }\r
-\r
- public SkipBMH(String pt, boolean ign)\r
- {\r
- this(pt, ign, 0);\r
- }\r
-\r
- public SkipBMH(String pt)\r
- {\r
- this(pt, false, 0);\r
- }\r
-\r
- public SkipBMH(String pt, boolean ign, int offset)\r
- {\r
- super(pt, ign, offset);\r
- for (int k = 0; k < MAX_CHAR; k++)\r
- {\r
- skip[k] = (char) src.length();\r
- }\r
-\r
- sm1 = src.length() - 1;\r
- x = src.charAt(sm1);\r
- uc = CaseMgr.toUpperCase(x);\r
- lc = CaseMgr.toLowerCase(x);\r
- tc = CaseMgr.toTitleCase(x);\r
-\r
- // We don't really want 65536 long arrays in skip[],\r
- // so we mask of the higher bits. This can be combined\r
- // with ignore case, so accounting for upper\r
- // case costs us nothing extra.\r
- for (int k = 0; k < src.length() - 1; k++)\r
- {\r
- char x_ = src.charAt(k);\r
- if (ign)\r
- {\r
- char uc_ = CaseMgr.toUpperCase(x_);\r
- char lc_ = CaseMgr.toLowerCase(x_);\r
- char tc_ = CaseMgr.toTitleCase(x_);\r
- skip[uc_ & (MAX_CHAR - 1)] = (char) (src.length() - k - 1);\r
- skip[lc_ & (MAX_CHAR - 1)] = (char) (src.length() - k - 1);\r
- skip[tc_ & (MAX_CHAR - 1)] = (char) (src.length() - k - 1);\r
- }\r
- else\r
- {\r
- skip[x_ & (MAX_CHAR - 1)] = (char) (src.length() - k - 1);\r
- }\r
- }\r
-\r
- // This trick can be found in the July issue of\r
- // C-T magazine. This makes the method a type of\r
- // "T-search."\r
- jump_ahead = src.length() - 1;\r
- for (int k = 0; k < src.length() - 1; k++)\r
- {\r
- char y = src.charAt(sm1 - k - 1);\r
- if (exact(y))\r
- {\r
- jump_ahead = k;\r
- break;\r
- }\r
- }\r
- }\r
-\r
- /** Set to true if you only want to compare two of the\r
- characters in the String. */\r
- final public int searchRegion(String s, int start, int end)\r
- {\r
- return find(s, start, end);\r
- }\r
-\r
- final public int searchFrom(String s, int start)\r
- {\r
- return find(s, start, s.length());\r
- }\r
-\r
- final public int search(String s)\r
- {\r
- return find(s, 0, s.length());\r
- }\r
-\r
- public int find(String s, int start, int end)\r
- {\r
- start += offset + sm1;\r
- int vend = min(s.length() - 1, end + sm1 + offset), k;\r
- int vend1 = vend - jump_ahead;\r
- if (ign)\r
- {\r
- for (k = start; k <= vend1; k += skip[s.charAt(k) & (MAX_CHAR - 1)])\r
- {\r
- // table look-up is expensive, avoid it if possible\r
- if (anyc(s.charAt(k)))\r
- {\r
- if (CaseMgr.regionMatches(src, ign, 0, s, k - sm1, sm1))\r
- {\r
- return k - sm1 - offset;\r
- }\r
- k += jump_ahead;\r
- }\r
- }\r
- for (; k <= vend; k += skip[s.charAt(k) & (MAX_CHAR - 1)])\r
- {\r
- // table look-up is expensive, avoid it if possible\r
- if (anyc(s.charAt(k)))\r
- {\r
- if (CaseMgr.regionMatches(src, ign, 0, s, k - sm1, sm1))\r
- {\r
- return k - sm1 - offset;\r
- }\r
- k += jump_ahead;\r
- if (k > vend)\r
- {\r
- return -1;\r
- }\r
- }\r
- }\r
- }\r
- else\r
- {\r
- for (k = start; k <= vend1; k += skip[s.charAt(k) & (MAX_CHAR - 1)])\r
- {\r
- // table look-up is expensive, avoid it if possible\r
- if (x == s.charAt(k))\r
- {\r
- //if(src.regionMatches(0,s,k-sm1,sm1))\r
- if (CaseMgr.regionMatches(src, false, 0, s, k - sm1, sm1))\r
- {\r
- return k - sm1 - offset;\r
- }\r
- k += jump_ahead;\r
- }\r
- }\r
- for (; k <= vend; k += skip[s.charAt(k) & (MAX_CHAR - 1)])\r
- {\r
- // table look-up is expensive, avoid it if possible\r
- if (x == s.charAt(k))\r
- {\r
- //if(src.regionMatches(0,s,k-sm1,sm1))\r
- if (CaseMgr.regionMatches(src, false, 0, s, k - sm1, sm1))\r
- {\r
- return k - sm1 - offset;\r
- }\r
- k += jump_ahead;\r
- if (k > vend)\r
- {\r
- return -1;\r
- }\r
- }\r
- }\r
- }\r
-\r
- return -1;\r
- }\r
-\r
- public int find(StringLike s, int start, int end)\r
- {\r
- if (s instanceof StringWrap)\r
- {\r
- return find(s.toString(), start, end);\r
- }\r
- start += offset + sm1;\r
- int vend = min(s.length() - 1, end + sm1 + offset), k;\r
- int vend1 = vend - jump_ahead;\r
- if (ign)\r
- {\r
- for (k = start; k <= vend1; k += skip[s.charAt(k) & (MAX_CHAR - 1)])\r
- {\r
- // table look-up is expensive, avoid it if possible\r
- if (anyc(s.charAt(k)))\r
- {\r
- if (CaseMgr.regionMatches(src, ign, 0, s, k - sm1, sm1))\r
- {\r
- return k - sm1 - offset;\r
- }\r
- k += jump_ahead;\r
- }\r
- }\r
- for (; k <= vend; k += skip[s.charAt(k) & (MAX_CHAR - 1)])\r
- {\r
- // table look-up is expensive, avoid it if possible\r
- if (anyc(s.charAt(k)))\r
- {\r
- if (CaseMgr.regionMatches(src, ign, 0, s, k - sm1, sm1))\r
- {\r
- return k - sm1 - offset;\r
- }\r
- k += jump_ahead;\r
- if (k > vend)\r
- {\r
- return -1;\r
- }\r
- }\r
- }\r
- }\r
- else\r
- {\r
- for (k = start; k <= vend1; k += skip[s.charAt(k) & (MAX_CHAR - 1)])\r
- {\r
- // table look-up is expensive, avoid it if possible\r
- if (x == s.charAt(k))\r
- {\r
- //if(src.regionMatches(0,s,k-sm1,sm1))\r
- if (CaseMgr.regionMatches(src, false, 0, s, k - sm1, sm1))\r
- {\r
- return k - sm1 - offset;\r
- }\r
- k += jump_ahead;\r
- }\r
- }\r
- for (; k <= vend; k += skip[s.charAt(k) & (MAX_CHAR - 1)])\r
- {\r
- // table look-up is expensive, avoid it if possible\r
- if (x == s.charAt(k))\r
- {\r
- //if(src.regionMatches(0,s,k-sm1,sm1))\r
- if (CaseMgr.regionMatches(src, false, 0, s, k - sm1, sm1))\r
- {\r
- return k - sm1 - offset;\r
- }\r
- k += jump_ahead;\r
- if (k > vend)\r
- {\r
- return -1;\r
- }\r
- }\r
- }\r
- }\r
-\r
- return -1;\r
- }\r
-}\r
+//
+// This software is now distributed according to
+// the Lesser Gnu Public License. Please see
+// http://www.gnu.org/copyleft/lesser.txt for
+// the details.
+// -- Happy Computing!
+//
+package com.stevesoft.pat;
+
+import com.stevesoft.pat.wrap.*;
+
+/**
+ * Like Skip, but implements a <a
+ * href="http://www.dcc.uchile.cl/~rbaeza/handbook/algs/7/713b.srch.p.html">
+ * Boyer-Moore-Horspool</a> type search method that has been modified to be
+ * more like a "T-search" (see the Michael Tamm''s article in <i>C'T, magazin
+ * fuer computer und technic</i>, August 97 p 292). Yet another important
+ * source of information for me was the <a
+ * href="http://www.go2net.com/people/paulp/deep/1997/05/14/"> Deep Magic</a>
+ * article on string searching. As of this writing, I can beat String's indexOf
+ * method in many cases.
+ *
+ * @see com.stevesoft.pat.Skip
+ * @see com.stevesoft.pat.Skip2
+ */
+public class SkipBMH extends Skip
+{
+ // This number could be 256, but I think it's
+ // big enough. Note, it must be a power of 2.
+ final int MAX_CHAR = 64;
+
+ final char[] skip = new char[MAX_CHAR];
+
+ int sm1;
+
+ int jump_ahead = 0;
+
+ char uc, lc, tc, x;
+
+ final boolean exact(char c)
+ {
+ return (ign && anyc(c)) || c == x;
+ }
+
+ final boolean anyc(char c)
+ {
+ return c == uc || c == lc || c == tc;
+ }
+
+ public SkipBMH(String pt, boolean ign)
+ {
+ this(pt, ign, 0);
+ }
+
+ public SkipBMH(String pt)
+ {
+ this(pt, false, 0);
+ }
+
+ public SkipBMH(String pt, boolean ign, int offset)
+ {
+ super(pt, ign, offset);
+ for (int k = 0; k < MAX_CHAR; k++)
+ {
+ skip[k] = (char) src.length();
+ }
+
+ sm1 = src.length() - 1;
+ x = src.charAt(sm1);
+ uc = CaseMgr.toUpperCaseC(x);
+ lc = CaseMgr.toLowerCaseC(x);
+ tc = CaseMgr.toTitleCaseC(x);
+
+ // We don't really want 65536 long arrays in skip[],
+ // so we mask of the higher bits. This can be combined
+ // with ignore case, so accounting for upper
+ // case costs us nothing extra.
+ for (int k = 0; k < src.length() - 1; k++)
+ {
+ char x_ = src.charAt(k);
+ if (ign)
+ {
+ char uc_ = CaseMgr.toUpperCaseC(x_);
+ char lc_ = CaseMgr.toLowerCaseC(x_);
+ char tc_ = CaseMgr.toTitleCaseC(x_);
+ skip[uc_ & (MAX_CHAR - 1)] = (char) (src.length() - k - 1);
+ skip[lc_ & (MAX_CHAR - 1)] = (char) (src.length() - k - 1);
+ skip[tc_ & (MAX_CHAR - 1)] = (char) (src.length() - k - 1);
+ }
+ else
+ {
+ skip[x_ & (MAX_CHAR - 1)] = (char) (src.length() - k - 1);
+ }
+ }
+
+ // This trick can be found in the July issue of
+ // C-T magazine. This makes the method a type of
+ // "T-search."
+ jump_ahead = src.length() - 1;
+ for (int k = 0; k < src.length() - 1; k++)
+ {
+ char y = src.charAt(sm1 - k - 1);
+ if (exact(y))
+ {
+ jump_ahead = k;
+ break;
+ }
+ }
+ }
+
+ /**
+ * Set to true if you only want to compare two of the characters in the
+ * String.
+ */
+ final public int searchRegion(String s, int start, int end)
+ {
+ return find(s, start, end);
+ }
+
+ final public int searchFrom(String s, int start)
+ {
+ return find(s, start, s.length());
+ }
+
+ final public int search(String s)
+ {
+ return find(s, 0, s.length());
+ }
+
+ public int find(String s, int start, int end)
+ {
+ start += offset + sm1;
+ int vend = min(s.length() - 1, end + sm1 + offset), k;
+ int vend1 = vend - jump_ahead;
+ if (ign)
+ {
+ for (k = start; k <= vend1; k += skip[s.charAt(k) & (MAX_CHAR - 1)])
+ {
+ // table look-up is expensive, avoid it if possible
+ if (anyc(s.charAt(k)))
+ {
+ if (CaseMgr.regionMatches2(src, ign, 0, s, k - sm1, sm1))
+ {
+ return k - sm1 - offset;
+ }
+ k += jump_ahead;
+ }
+ }
+ for (; k <= vend; k += skip[s.charAt(k) & (MAX_CHAR - 1)])
+ {
+ // table look-up is expensive, avoid it if possible
+ if (anyc(s.charAt(k)))
+ {
+ if (CaseMgr.regionMatches2(src, ign, 0, s, k - sm1, sm1))
+ {
+ return k - sm1 - offset;
+ }
+ k += jump_ahead;
+ if (k > vend)
+ {
+ return -1;
+ }
+ }
+ }
+ }
+ else
+ {
+ for (k = start; k <= vend1; k += skip[s.charAt(k) & (MAX_CHAR - 1)])
+ {
+ // table look-up is expensive, avoid it if possible
+ if (x == s.charAt(k))
+ {
+ // if(src.regionMatches(0,s,k-sm1,sm1))
+ if (CaseMgr.regionMatches2(src, false, 0, s, k - sm1, sm1))
+ {
+ return k - sm1 - offset;
+ }
+ k += jump_ahead;
+ }
+ }
+ for (; k <= vend; k += skip[s.charAt(k) & (MAX_CHAR - 1)])
+ {
+ // table look-up is expensive, avoid it if possible
+ if (x == s.charAt(k))
+ {
+ // if(src.regionMatches(0,s,k-sm1,sm1))
+ if (CaseMgr.regionMatches2(src, false, 0, s, k - sm1, sm1))
+ {
+ return k - sm1 - offset;
+ }
+ k += jump_ahead;
+ if (k > vend)
+ {
+ return -1;
+ }
+ }
+ }
+ }
+
+ return -1;
+ }
+
+ public int find(StringLike s, int start, int end)
+ {
+ if (s instanceof StringWrap)
+ {
+ return find(s.toString(), start, end);
+ }
+ start += offset + sm1;
+ int vend = min(s.length() - 1, end + sm1 + offset), k;
+ int vend1 = vend - jump_ahead;
+ if (ign)
+ {
+ for (k = start; k <= vend1; k += skip[s.charAt(k) & (MAX_CHAR - 1)])
+ {
+ // table look-up is expensive, avoid it if possible
+ if (anyc(s.charAt(k)))
+ {
+ if (CaseMgr.regionMatches(src, ign, 0, s, k - sm1, sm1))
+ {
+ return k - sm1 - offset;
+ }
+ k += jump_ahead;
+ }
+ }
+ for (; k <= vend; k += skip[s.charAt(k) & (MAX_CHAR - 1)])
+ {
+ // table look-up is expensive, avoid it if possible
+ if (anyc(s.charAt(k)))
+ {
+ if (CaseMgr.regionMatches(src, ign, 0, s, k - sm1, sm1))
+ {
+ return k - sm1 - offset;
+ }
+ k += jump_ahead;
+ if (k > vend)
+ {
+ return -1;
+ }
+ }
+ }
+ }
+ else
+ {
+ for (k = start; k <= vend1; k += skip[s.charAt(k) & (MAX_CHAR - 1)])
+ {
+ // table look-up is expensive, avoid it if possible
+ if (x == s.charAt(k))
+ {
+ // if(src.regionMatches(0,s,k-sm1,sm1))
+ if (CaseMgr.regionMatches(src, false, 0, s, k - sm1, sm1))
+ {
+ return k - sm1 - offset;
+ }
+ k += jump_ahead;
+ }
+ }
+ for (; k <= vend; k += skip[s.charAt(k) & (MAX_CHAR - 1)])
+ {
+ // table look-up is expensive, avoid it if possible
+ if (x == s.charAt(k))
+ {
+ // if(src.regionMatches(0,s,k-sm1,sm1))
+ if (CaseMgr.regionMatches(src, false, 0, s, k - sm1, sm1))
+ {
+ return k - sm1 - offset;
+ }
+ k += jump_ahead;
+ if (k > vend)
+ {
+ return -1;
+ }
+ }
+ }
+ }
+
+ return -1;
+ }
+}
git://source.jalview.org / jalview.git / blobdiff