summaryrefslogtreecommitdiffstats
path: root/src/shlex.cc
diff options
context:
space:
mode:
Diffstat (limited to '')
-rw-r--r--src/shlex.cc312
1 files changed, 261 insertions, 51 deletions
diff --git a/src/shlex.cc b/src/shlex.cc
index 8da44bb..d33924e 100644
--- a/src/shlex.cc
+++ b/src/shlex.cc
@@ -34,42 +34,70 @@
#endif
#include "config.h"
+#include "pcrepp/pcre2pp.hh"
#include "shlex.hh"
-bool
-shlex::tokenize(string_fragment& cap_out, shlex_token_t& token_out)
+using namespace lnav::roles::literals;
+
+std::string
+shlex::escape(std::string s)
+{
+ static const auto SH_CHARS = lnav::pcre2pp::code::from_const("'");
+
+ return SH_CHARS.replace(s, "\\'");
+}
+
+attr_line_t
+shlex::to_attr_line(const shlex::tokenize_error_t& te) const
+{
+ return attr_line_t()
+ .append(string_fragment::from_bytes(this->s_str, this->s_len))
+ .append("\n")
+ .pad_to(te.te_source.sf_begin)
+ .append("^"_snippet_border);
+}
+
+Result<shlex::tokenize_result_t, shlex::tokenize_error_t>
+shlex::tokenize()
{
+ tokenize_result_t retval;
+
+ retval.tr_frag.sf_string = this->s_str;
while (this->s_index < this->s_len) {
switch (this->s_str[this->s_index]) {
case '\\':
- cap_out.sf_begin = this->s_index;
+ retval.tr_frag.sf_begin = this->s_index;
if (this->s_index + 1 < this->s_len) {
- token_out = shlex_token_t::ST_ESCAPE;
+ retval.tr_token = shlex_token_t::escape;
this->s_index += 2;
- cap_out.sf_end = this->s_index;
+ retval.tr_frag.sf_end = this->s_index;
} else {
this->s_index += 1;
- cap_out.sf_end = this->s_index;
- token_out = shlex_token_t::ST_ERROR;
+ retval.tr_frag.sf_end = this->s_index;
+
+ return Err(tokenize_error_t{
+ "invalid escape",
+ retval.tr_frag,
+ });
}
- return true;
+ return Ok(retval);
case '\"':
if (!this->s_ignore_quotes) {
switch (this->s_state) {
case state_t::STATE_NORMAL:
- cap_out.sf_begin = this->s_index;
+ retval.tr_frag.sf_begin = this->s_index;
this->s_index += 1;
- cap_out.sf_end = this->s_index;
- token_out = shlex_token_t::ST_DOUBLE_QUOTE_START;
+ retval.tr_frag.sf_end = this->s_index;
+ retval.tr_token = shlex_token_t::double_quote_start;
this->s_state = state_t::STATE_IN_DOUBLE_QUOTE;
- return true;
+ return Ok(retval);
case state_t::STATE_IN_DOUBLE_QUOTE:
- cap_out.sf_begin = this->s_index;
+ retval.tr_frag.sf_begin = this->s_index;
this->s_index += 1;
- cap_out.sf_end = this->s_index;
- token_out = shlex_token_t::ST_DOUBLE_QUOTE_END;
+ retval.tr_frag.sf_end = this->s_index;
+ retval.tr_token = shlex_token_t::double_quote_end;
this->s_state = state_t::STATE_NORMAL;
- return true;
+ return Ok(retval);
default:
break;
}
@@ -79,19 +107,19 @@ shlex::tokenize(string_fragment& cap_out, shlex_token_t& token_out)
if (!this->s_ignore_quotes) {
switch (this->s_state) {
case state_t::STATE_NORMAL:
- cap_out.sf_begin = this->s_index;
+ retval.tr_frag.sf_begin = this->s_index;
this->s_index += 1;
- cap_out.sf_end = this->s_index;
- token_out = shlex_token_t::ST_SINGLE_QUOTE_START;
+ retval.tr_frag.sf_end = this->s_index;
+ retval.tr_token = shlex_token_t::single_quote_start;
this->s_state = state_t::STATE_IN_SINGLE_QUOTE;
- return true;
+ return Ok(retval);
case state_t::STATE_IN_SINGLE_QUOTE:
- cap_out.sf_begin = this->s_index;
+ retval.tr_frag.sf_begin = this->s_index;
this->s_index += 1;
- cap_out.sf_end = this->s_index;
- token_out = shlex_token_t::ST_SINGLE_QUOTE_END;
+ retval.tr_frag.sf_end = this->s_index;
+ retval.tr_token = shlex_token_t::single_quote_end;
this->s_state = state_t::STATE_NORMAL;
- return true;
+ return Ok(retval);
default:
break;
}
@@ -100,9 +128,10 @@ shlex::tokenize(string_fragment& cap_out, shlex_token_t& token_out)
case '$':
switch (this->s_state) {
case state_t::STATE_NORMAL:
- case state_t::STATE_IN_DOUBLE_QUOTE:
- this->scan_variable_ref(cap_out, token_out);
- return true;
+ case state_t::STATE_IN_DOUBLE_QUOTE: {
+ auto rc = TRY(this->scan_variable_ref());
+ return Ok(rc);
+ }
default:
break;
}
@@ -110,7 +139,7 @@ shlex::tokenize(string_fragment& cap_out, shlex_token_t& token_out)
case '~':
switch (this->s_state) {
case state_t::STATE_NORMAL:
- cap_out.sf_begin = this->s_index;
+ retval.tr_frag.sf_begin = this->s_index;
this->s_index += 1;
while (this->s_index < this->s_len
&& (isalnum(this->s_str[this->s_index])
@@ -119,9 +148,9 @@ shlex::tokenize(string_fragment& cap_out, shlex_token_t& token_out)
{
this->s_index += 1;
}
- cap_out.sf_end = this->s_index;
- token_out = shlex_token_t::ST_TILDE;
- return true;
+ retval.tr_frag.sf_end = this->s_index;
+ retval.tr_token = shlex_token_t::tilde;
+ return Ok(retval);
default:
break;
}
@@ -130,13 +159,15 @@ shlex::tokenize(string_fragment& cap_out, shlex_token_t& token_out)
case '\t':
switch (this->s_state) {
case state_t::STATE_NORMAL:
- cap_out.sf_begin = this->s_index;
- while (isspace(this->s_str[this->s_index])) {
+ retval.tr_frag.sf_begin = this->s_index;
+ while (this->s_index < this->s_len
+ && isspace(this->s_str[this->s_index]))
+ {
this->s_index += 1;
}
- cap_out.sf_end = this->s_index;
- token_out = shlex_token_t::ST_WHITESPACE;
- return true;
+ retval.tr_frag.sf_end = this->s_index;
+ retval.tr_token = shlex_token_t::whitespace;
+ return Ok(retval);
default:
break;
}
@@ -148,29 +179,47 @@ shlex::tokenize(string_fragment& cap_out, shlex_token_t& token_out)
this->s_index += 1;
}
- return false;
+ if (this->s_state != state_t::STATE_NORMAL) {
+ retval.tr_frag.sf_begin = this->s_index;
+ retval.tr_frag.sf_end = this->s_len;
+ return Err(tokenize_error_t{
+ "non-terminated string",
+ retval.tr_frag,
+ });
+ }
+
+ retval.tr_frag.sf_begin = this->s_len;
+ retval.tr_frag.sf_end = this->s_len;
+ retval.tr_token = shlex_token_t::eof;
+ return Ok(retval);
}
-void
-shlex::scan_variable_ref(string_fragment& cap_out, shlex_token_t& token_out)
+Result<shlex::tokenize_result_t, shlex::tokenize_error_t>
+shlex::scan_variable_ref()
{
- cap_out.sf_begin = this->s_index;
+ tokenize_result_t retval;
+
+ retval.tr_frag.sf_string = this->s_str;
+
+ retval.tr_frag.sf_begin = this->s_index;
this->s_index += 1;
if (this->s_index >= this->s_len) {
- cap_out.sf_end = this->s_index;
- token_out = shlex_token_t::ST_ERROR;
- return;
+ retval.tr_frag.sf_end = this->s_index;
+ return Err(tokenize_error_t{
+ "invalid variable reference",
+ retval.tr_frag,
+ });
}
if (this->s_str[this->s_index] == '{') {
- token_out = shlex_token_t::ST_QUOTED_VARIABLE_REF;
+ retval.tr_token = shlex_token_t::quoted_variable_ref;
this->s_index += 1;
} else {
- token_out = shlex_token_t::ST_VARIABLE_REF;
+ retval.tr_token = shlex_token_t::variable_ref;
}
while (this->s_index < this->s_len) {
- if (token_out == shlex_token_t::ST_VARIABLE_REF) {
+ if (retval.tr_token == shlex_token_t::variable_ref) {
if (isalnum(this->s_str[this->s_index])
|| this->s_str[this->s_index] == '#'
|| this->s_str[this->s_index] == '_')
@@ -188,14 +237,19 @@ shlex::scan_variable_ref(string_fragment& cap_out, shlex_token_t& token_out)
}
}
- cap_out.sf_end = this->s_index;
- if (token_out == shlex_token_t::ST_QUOTED_VARIABLE_REF
+ retval.tr_frag.sf_end = this->s_index;
+ if (retval.tr_token == shlex_token_t::quoted_variable_ref
&& this->s_str[this->s_index - 1] != '}')
{
- cap_out.sf_begin += 1;
- cap_out.sf_end = cap_out.sf_begin + 1;
- token_out = shlex_token_t::ST_ERROR;
+ retval.tr_frag.sf_begin += 1;
+ retval.tr_frag.sf_end = retval.tr_frag.sf_begin + 1;
+ return Err(tokenize_error_t{
+ "missing closing curly-brace in variable reference",
+ retval.tr_frag,
+ });
}
+
+ return Ok(retval);
}
void
@@ -216,3 +270,159 @@ shlex::resolve_home_dir(std::string& result, string_fragment cap) const
}
}
}
+
+bool
+shlex::eval(std::string& result, const scoped_resolver& vars)
+{
+ result.clear();
+
+ int last_index = 0;
+ bool done = false;
+
+ while (!done) {
+ auto tokenize_res = this->tokenize();
+ if (tokenize_res.isErr()) {
+ return false;
+ }
+ auto token = tokenize_res.unwrap();
+
+ result.append(&this->s_str[last_index],
+ token.tr_frag.sf_begin - last_index);
+ switch (token.tr_token) {
+ case shlex_token_t::eof:
+ done = true;
+ break;
+ case shlex_token_t::escape:
+ result.append(1, this->s_str[token.tr_frag.sf_begin + 1]);
+ break;
+ case shlex_token_t::whitespace:
+ result.append(&this->s_str[token.tr_frag.sf_begin],
+ token.tr_frag.length());
+ break;
+ case shlex_token_t::variable_ref:
+ case shlex_token_t::quoted_variable_ref: {
+ int extra = token.tr_token == shlex_token_t::variable_ref ? 0
+ : 1;
+ const std::string var_name(
+ &this->s_str[token.tr_frag.sf_begin + 1 + extra],
+ token.tr_frag.length() - 1 - extra * 2);
+ auto local_var = vars.find(var_name);
+ const char* var_value = getenv(var_name.c_str());
+
+ if (local_var != vars.end()) {
+ result.append(fmt::to_string(local_var->second));
+ } else if (var_value != nullptr) {
+ result.append(var_value);
+ }
+ break;
+ }
+ case shlex_token_t::tilde:
+ this->resolve_home_dir(result, token.tr_frag);
+ break;
+ case shlex_token_t::double_quote_start:
+ case shlex_token_t::double_quote_end:
+ result.append("\"");
+ break;
+ case shlex_token_t::single_quote_start:
+ case shlex_token_t::single_quote_end:
+ result.append("'");
+ break;
+ default:
+ break;
+ }
+ last_index = token.tr_frag.sf_end;
+ }
+
+ result.append(&this->s_str[last_index], this->s_len - last_index);
+
+ return true;
+}
+
+Result<std::vector<shlex::split_element_t>, shlex::tokenize_error_t>
+shlex::split(const scoped_resolver& vars)
+{
+ std::vector<split_element_t> retval;
+ int last_index = 0;
+ bool start_new = true;
+ bool done = false;
+
+ while (this->s_index < this->s_len && isspace(this->s_str[this->s_index])) {
+ this->s_index += 1;
+ }
+ if (this->s_index == this->s_len) {
+ return Ok(retval);
+ }
+ while (!done) {
+ auto tokenize_res = TRY(this->tokenize());
+
+ if (start_new) {
+ if (last_index < this->s_len) {
+ retval.emplace_back(split_element_t{
+ string_fragment::from_byte_range(
+ this->s_str, last_index, tokenize_res.tr_frag.sf_begin),
+ "",
+ });
+ }
+ start_new = false;
+ } else if (tokenize_res.tr_token != shlex_token_t::whitespace) {
+ retval.back().se_origin.sf_end = tokenize_res.tr_frag.sf_end;
+ } else {
+ retval.back().se_origin.sf_end = tokenize_res.tr_frag.sf_begin;
+ }
+ retval.back().se_value.append(
+ &this->s_str[last_index],
+ tokenize_res.tr_frag.sf_begin - last_index);
+ switch (tokenize_res.tr_token) {
+ case shlex_token_t::eof:
+ done = true;
+ break;
+ case shlex_token_t::escape:
+ retval.back().se_value.append(
+ 1, this->s_str[tokenize_res.tr_frag.sf_begin + 1]);
+ break;
+ case shlex_token_t::whitespace:
+ start_new = true;
+ break;
+ case shlex_token_t::variable_ref:
+ case shlex_token_t::quoted_variable_ref: {
+ int extra = tokenize_res.tr_token == shlex_token_t::variable_ref
+ ? 0
+ : 1;
+ std::string var_name(
+ &this->s_str[tokenize_res.tr_frag.sf_begin + 1 + extra],
+ tokenize_res.tr_frag.length() - 1 - extra * 2);
+ auto local_var = vars.find(var_name);
+ const char* var_value = getenv(var_name.c_str());
+
+ if (local_var != vars.end()) {
+ retval.back().se_value.append(
+ fmt::to_string(local_var->second));
+ } else if (var_value != nullptr) {
+ retval.back().se_value.append(var_value);
+ }
+ break;
+ }
+ case shlex_token_t::tilde:
+ this->resolve_home_dir(retval.back().se_value,
+ tokenize_res.tr_frag);
+ break;
+ default:
+ break;
+ }
+ last_index = tokenize_res.tr_frag.sf_end;
+ }
+
+ if (last_index < this->s_len) {
+ if (start_new || retval.empty()) {
+ retval.emplace_back(split_element_t{
+ string_fragment::from_byte_range(
+ this->s_str, last_index, this->s_len),
+ "",
+ });
+ }
+ retval.back().se_value.append(&this->s_str[last_index],
+ this->s_len - last_index);
+ }
+
+ return Ok(retval);
+}
generated by cgit v1.2.3 (git 2.39.1) at 2024-11-11 02:14:14 +0000