尽管我不得不质疑这个方案的实用性——我倾向于支持那些在没有达到预期时大声抱怨的系统，而不是像这样更加松散、笨拙的系统——我相信下面的脚本会做你想做的事。
操作理论：它预先读取所有内容，然后一举将其输出全部发出。
假设您将文件命名injector为injector input.conf super.conf.
#!/usr/bin/env awk -f
#
# Expects to be called with two files. First is the content to inject,
# second is the file to inject into.
FNR == 1 {
# This switches from "read replacement content" to "read template"
# at the boundary between reading the first and second files. This
# will of course do something suprising if you pass more than two
# files.
readReplacement = !readReplacement;
}
# Read a line of replacement content.
readReplacement {
rCount++;
replacement[rCount] = $0;
next;
}
# Read a line of template content.
{
tCount++;
template[tCount] = $0;
}
# Note the beginning of the replacement area.
/^#blockbegin$/ {
beginAt = tCount;
}
# Note the end of the replacement area.
/^#blockend$/ {
endAt = tCount;
}
# Finished reading everything. Process it all.
END {
if (beginAt && endAt) {
# Both beginning and ending markers were found; replace what's
# in the middle of them.
emitTemplate(1, beginAt);
emitReplacement();
emitTemplate(endAt, tCount);
} else {
# Didn't find both markers; just append.
emitTemplate(1, tCount);
emitReplacement();
}
}
# Emit the indicated portion of the template to stdout.
function emitTemplate(from, to) {
for (i = from; i <= to; i++) {
print template[i];
}
}
# Emit the replacement text to stdout.
function emitReplacement() {
for (i = 1; i <= rCount; i++) {
print replacement[i];
}
}