From 39e54b9167feb6bbc9994ed88ac9dda64be6246f Mon Sep 17 00:00:00 2001
From: Diego Frias <mail@dzfrias.dev>
Date: Sat, 5 Jul 2025 12:40:04 -0700
Subject: [PATCH] ICU-23156 Implement `icupkg --keep` option

---
 icu4c/source/tools/icupkg/icupkg.cpp    | 23 ++++++++++++++++-
 icu4c/source/tools/toolutil/package.cpp | 33 +++++++++++++++++++++++++
 icu4c/source/tools/toolutil/package.h   |  4 +++
 3 files changed, 59 insertions(+), 1 deletion(-)

diff --git a/icu4c/source/tools/icupkg/icupkg.cpp b/icu4c/source/tools/icupkg/icupkg.cpp
index 3da4af745786..cc65b4f8a24f 100644
--- a/icu4c/source/tools/icupkg/icupkg.cpp
+++ b/icu4c/source/tools/icupkg/icupkg.cpp
@@ -57,7 +57,7 @@ printUsage(const char *pname, UBool isHelp) {
 
     fprintf(where,
             "%csage: %s [-h|-?|--help ] [-tl|-tb|-te] [-c] [-C comment]\n"
-            "\t[-a list] [-r list] [-x list] [-l [-o outputListFileName]]\n"
+            "\t[-a list] [-r list] [-x list] [-k list] [-l [-o outputListFileName]]\n"
             "\t[-s path] [-d path] [-w] [-m mode]\n"
             "\t[--ignore-deps]\n"
             "\t[--auto_toc_prefix] [--auto_toc_prefix_with_type] [--toc_prefix]\n"
@@ -108,6 +108,7 @@ printUsage(const char *pname, UBool isHelp) {
             "\t-a list or --add list      add items to the package\n"
             "\t-r list or --remove list   remove items from the package\n"
             "\t-x list or --extract list  extract items from the package\n"
+            "\t-k list or --keep list     keep only certain items from the package\n"
             "\tThe list can be a single item's filename,\n"
             "\tor a .txt filename with a list of item filenames,\n"
             "\tor an ICU .dat package filename.\n");
@@ -208,6 +209,7 @@ static UOption options[]={
     UOPTION_DEF("add", 'a', UOPT_REQUIRES_ARG),
     UOPTION_DEF("remove", 'r', UOPT_REQUIRES_ARG),
     UOPTION_DEF("extract", 'x', UOPT_REQUIRES_ARG),
+    UOPTION_DEF("keep", 'k', UOPT_REQUIRES_ARG),
 
     UOPTION_DEF("list", 'l', UOPT_NO_ARG),
     UOPTION_DEF("outlist", 'o', UOPT_REQUIRES_ARG),
@@ -237,6 +239,7 @@ enum {
     OPT_ADD_LIST,
     OPT_REMOVE_LIST,
     OPT_EXTRACT_LIST,
+    OPT_KEEP_LIST,
 
     OPT_LIST_ITEMS,
     OPT_LIST_FILE,
@@ -400,6 +403,7 @@ main(int argc, char *argv[]) {
             options[OPT_REMOVE_LIST].doesOccur ||
             options[OPT_ADD_LIST].doesOccur ||
             options[OPT_EXTRACT_LIST].doesOccur ||
+            options[OPT_KEEP_LIST].doesOccur ||
             options[OPT_LIST_ITEMS].doesOccur
         ) {
             printUsage(pname, false);
@@ -487,6 +491,23 @@ main(int argc, char *argv[]) {
         }
     }
 
+    /* keep items */
+    if(options[OPT_KEEP_LIST].doesOccur) {
+        listPkg=new Package();
+        if(listPkg==nullptr) {
+            fprintf(stderr, "icupkg: not enough memory\n");
+            exit(U_MEMORY_ALLOCATION_ERROR);
+        }
+        if(readList(nullptr, options[OPT_KEEP_LIST].value, false, listPkg)) {
+            pkg->keepItems(*listPkg);
+            delete listPkg;
+            isModified=true;
+        } else {
+            printUsage(pname, false);
+            return U_ILLEGAL_ARGUMENT_ERROR;
+        }
+    }
+
     /* list items */
     if(options[OPT_LIST_ITEMS].doesOccur) {
         int32_t i;
diff --git a/icu4c/source/tools/toolutil/package.cpp b/icu4c/source/tools/toolutil/package.cpp
index e58c6648b647..7b827036d204 100644
--- a/icu4c/source/tools/toolutil/package.cpp
+++ b/icu4c/source/tools/toolutil/package.cpp
@@ -431,6 +431,18 @@ Package::~Package() {
     uprv_free((void*)items);
 }
 
+void
+Package::clear() {
+    for(int32_t idx=0; idx<itemCount; ++idx) {
+        if(items[idx].isDataOwned) {
+            uprv_free(items[idx].data);
+        }
+    }
+    memset(items, 0, itemCount*sizeof(Item));
+    itemCount=0;
+    findNextIndex=-1;
+}
+
 void
 Package::setPrefix(const char *p) {
     if(strlen(p)>=sizeof(pkgPrefix)) {
@@ -1206,6 +1218,27 @@ Package::extractItems(const char *filesPath, const Package &listPkg, char outTyp
     }
 }
 
+void
+Package::keepItems(const Package &listPkg) {
+    Package *keepPkg=new Package();
+
+    const Item *pItem;
+    int32_t i;
+    for(pItem=listPkg.items, i=0; i<listPkg.itemCount; ++pItem, ++i) {
+        int32_t idx;
+        findItems(pItem->name);
+        while((idx=findNextItem())>=0) {
+            const Item *item=getItem(idx);
+            keepPkg->addItem(item->name, item->data, item->length, false, item->type);
+        }
+    }
+
+    clear();
+    addItems(*keepPkg);
+
+    delete keepPkg;
+}
+
 int32_t
 Package::getItemCount() const {
     return itemCount;
diff --git a/icu4c/source/tools/toolutil/package.h b/icu4c/source/tools/toolutil/package.h
index ea60c13a74a5..023ffde69665 100644
--- a/icu4c/source/tools/toolutil/package.h
+++ b/icu4c/source/tools/toolutil/package.h
@@ -53,6 +53,8 @@ class U_TOOLUTIL_API Package {
     /* Destructor. */
     ~Package();
 
+    void clear();
+
     /**
      * Uses the prefix of the first entry of the package in readPackage(),
      * rather than the package basename.
@@ -124,6 +126,8 @@ class U_TOOLUTIL_API Package {
     /* This variant extracts an item to a specific filename. */
     void extractItem(const char *filesPath, const char *outName, int32_t itemIndex, char outType);
 
+    void keepItems(const Package &listPkg);
+
     int32_t getItemCount() const;
     const Item *getItem(int32_t idx) const;