fixed UTF-8 print in windows

qinwf · Nov 6, 2014 · 295de8b · 295de8b
1 parent ddd75e6
commit 295de8b
Show file tree

Hide file tree

Showing 6 changed files with 46 additions and 46 deletions.
diff --git a/.gitignore b/.gitignore
@@ -9,3 +9,4 @@ packrat/lib*/
 .Rprofile
 *src-i386/
 *src-x64/
+*.txt
diff --git a/R/keywords.R b/R/keywords.R
@@ -34,7 +34,10 @@ keywords <- function(code, jiebar) {
 }
 
 keyl <- function(code, jiebar, encoding) {
-
+  if (.Platform$OS.type == "windows"){
+    old.locale <- Sys.getlocale("LC_CTYPE")
+    Sys.setlocale(category = "LC_CTYPE", locale = "chs")
+  }
   input.r <- file(code, open = "r")
   OUT <- FALSE
   tryCatch({
@@ -52,6 +55,9 @@ keyl <- function(code, jiebar, encoding) {
     return(out.lines)
   }, finally = {
     try(close(input.r), silent = TRUE)
+    if (.Platform$OS.type == "windows"){
+      Sys.setlocale(category = "LC_CTYPE", locale = old.locale)
+    }
   })
 }
 
@@ -66,8 +72,6 @@ keyw <- function(code, jiebar) {
     result <- jiebar$worker$tag(code)
 
   if (.Platform$OS.type == "windows") {
-
-    return(iconv(result, "UTF-8", "GBK"))
-  }
+    Encoding(result)<-"UTF-8"}
   return(result)
 } 
diff --git a/R/segment.R b/R/segment.R
@@ -59,13 +59,14 @@ segment <- function(code, jiebar) {
 }
 
 cutl <- function(code, jiebar, symbol, lines, output, encoding, write_file,FILESMODE) {
+  if (.Platform$OS.type == "windows"){
+    old.locale <- Sys.getlocale("LC_CTYPE")
+    Sys.setlocale(category = "LC_CTYPE", locale = "chs")
+  }
   nlines <- lines
   input.r <- file(code, open = "r")
   if(write_file==T){
-    if (.Platform$OS.type == "windows"){
-      old.locale <- Sys.getlocale("LC_CTYPE")
-      Sys.setlocale(category = "LC_CTYPE", locale = "chs")
-    }
+
     if (.Platform$OS.type == "windows") {
       output.w <- file(output, open = "ab", encoding = "UTF-8")
     } else {
@@ -127,6 +128,9 @@ cutl <- function(code, jiebar, symbol, lines, output, encoding, write_file,FILES
     }
     , finally = {
       try(close(input.r), silent = TRUE)
+      if (.Platform$OS.type == "windows"){
+        Sys.setlocale(category = "LC_CTYPE", locale = old.locale)
+      }
     })
     return(result)
   }

diff --git a/R/simhash.R b/R/simhash.R
@@ -36,7 +36,10 @@ simhash <- function(code, jiebar) {
 }
 
 simhashl <- function(code, jiebar, encoding) {
-
+  if (.Platform$OS.type == "windows"){
+    old.locale <- Sys.getlocale("LC_CTYPE")
+    Sys.setlocale(category = "LC_CTYPE", locale = "chs")
+  }
   input.r <- file(code, open = "r")
   OUT <- FALSE
   tryCatch({
@@ -53,6 +56,9 @@ simhashl <- function(code, jiebar, encoding) {
     return(out.lines)
   }, finally = {
     try(close(input.r), silent = TRUE)
+    if (.Platform$OS.type == "windows"){
+      Sys.setlocale(category = "LC_CTYPE", locale = old.locale)
+    }
   })
 }
 
@@ -67,9 +73,7 @@ simhashw <- function(code, jiebar) {
     result <- jiebar$worker$simhash(code,jiebar$topn)
 
   if (.Platform$OS.type == "windows") {
-    result$keyword<-iconv(result$keyword, "UTF-8", "GBK")
-
-    return(result)
+    Encoding(result$keyword)<-"UTF-8"
   }
   return(result)
 } 
@@ -111,15 +115,18 @@ distance<-function(codel,coder,jiebar){
   }
   result<-jiebar$worker$distance(codel,coder,jiebar$topn)
   if (.Platform$OS.type == "windows") {
-    result$lhs<-iconv(result$lhs, "UTF-8", "GBK")
-    result$rhs<-iconv(result$rhs, "UTF-8", "GBK")
+    Encoding(result$rhs)<-"UTF-8"
+    Encoding(result$lhs)<-"UTF-8"
   }
   result  
 }
 
 
 distancel <- function(code, jiebar, encoding) {
-
+  if (.Platform$OS.type == "windows"){
+    old.locale <- Sys.getlocale("LC_CTYPE")
+    Sys.setlocale(category = "LC_CTYPE", locale = "chs")
+  }
   input.r <- file(code, open = "r")
   OUT <- FALSE
   tryCatch({
@@ -140,5 +147,8 @@ distancel <- function(code, jiebar, encoding) {
     return(tmp.lines)
   }, finally = {
     try(close(input.r), silent = TRUE)
+    if (.Platform$OS.type == "windows"){
+      Sys.setlocale(category = "LC_CTYPE", locale = old.locale)
+    }
   })
 }
diff --git a/R/tagger.R b/R/tagger.R
@@ -42,6 +42,10 @@ tag<- function(code, jiebar) {
 }
 
 tagl <- function(code, jiebar, symbol, lines, output, encoding, write_file,FILESMODE) {
+  if (.Platform$OS.type == "windows"){
+    old.locale <- Sys.getlocale("LC_CTYPE")
+    Sys.setlocale(category = "LC_CTYPE", locale = "chs")
+  }
   nlines <- lines
   input.r <- file(code, open = "r")
   if(write_file==T){
@@ -76,6 +80,9 @@ tagl <- function(code, jiebar, symbol, lines, output, encoding, write_file,FILES
     , finally = {
       try(close(input.r), silent = TRUE)
       try(close(output.w), silent = TRUE)
+      if (.Platform$OS.type == "windows"){
+        Sys.setlocale(category = "LC_CTYPE", locale = old.locale)
+      }
     })
     OUT <- TRUE
     cat(paste("Output file: ", output, "\n"))
@@ -103,6 +110,9 @@ tagl <- function(code, jiebar, symbol, lines, output, encoding, write_file,FILES
     }
     , finally = {
       try(close(input.r), silent = TRUE)
+      if (.Platform$OS.type == "windows"){
+        Sys.setlocale(category = "LC_CTYPE", locale = old.locale)
+      }
     })
 
     return(result)
@@ -126,9 +136,7 @@ tagw <- function(code, jiebar,  symbol, FILESMODE) {
   if (symbol == F && FILESMODE  ==F) {
     result <- grep("[^[:space:]]", result, value = T)
   }
-  if (.Platform$OS.type == "windows" && FILESMODE == F) {
-
-    return(iconv(result, "UTF-8", "GBK"))
-  }
+  if (.Platform$OS.type == "windows") {
+    Encoding(result)<-"UTF-8"}
   return(result)
 } 
diff --git a/RR.txt b/RR.txt