Intermezzo: make my search history computable

So, how can one go from the search history above to something, that can be saved and used programmatically. First off, it's quite easy to drag the history entries from Safari into a file folder and end up with a bunch of .webloc files like in the above list. But how do one go from those files to the actual URLs?

In[1]:= SetDirectory["~/2025"];

In[2]:= FileNames[]//First
Out[2]= 21st century bjarne stoustrup linkedin at DuckDuckGo 2.webloc

In[3]:= FilePrint@%
         bplist00Ñ\.01\.02SURL_\.10Fhttps://duckduckgo.com/?q=21st+century+bjarne+stoustrup+linkedin&t=osx\.08\.0b\.0f

In[4]:= Read[%2]
During evaluation of In[4]:= $CharacterEncoding::utf8: The byte sequence {209,1} could not be interpreted as a character in the UTF-8 character encoding.
During evaluation of In[4]:= $CharacterEncoding::utf8: The byte sequence {209,1} could not be interpreted as a character in the UTF-8 character encoding.
During evaluation of In[4]:= Read::readt: "Invalid input found when reading \!\(\*RowBox[{\"\\\"bplist00Ñ\.02SURL_\.10Fhttps://duckduckgo.com/?q=21st+century+bjarne+stoustrup+linkedin&t=osx\\b\.0b\.0f\\\\\\\\:0000\\\\\\\\:0000\\\\\\\\:0000\\\\\\\\:0000\\\\\\\\:0000\\\\\\\\:0000\.01\\\\\\\\:0000\\\\\\\\:0000\\\\\\\\:0000\\\\\\\\:0000\\\\\\\\:0000\\\\\\\\:0000\\\\\\\\:0000\.03\\\\\\\\:0000\\\\\\\\:0000\\\\\\\\:0000\\\\\\\\:0000\\\\\\\\:0000\\\\\\\\:0000\\\\\\\\:0000\\\\\\\\:0000\\\\\\\\:0000\\\\\\\\:0000\\\\\\\\:0000\\\\\\\\:0000\\\\\\\\:0000\\\\\\\\:0000\\\\\\\\:0000X\\\\n\\\"\"}]\) from \!\(\*RowBox[{\"\\\"21st century bjarne stoustrup linkedin at DuckDuckGo 2.webloc\\\"\"}]\)."
Out[4]= $Failed
        

Alas, the files are in a binary format. Again, off to the web to learn about this format. I quickly realised that this off course was Apple property files in binary format. Years ago I worked a bit with such files, but back then they were in XML format. So, when I knew the file type, it became easy to convert them to XML:

>: for f in *.webloc
do
   plutil -convert xml1 $f
done

I could now easy read the XML files and transform their content into a simple list of URLs and put that list on the clipboard for pasting it into this HTML file:

RunProcess["pbcopy","StandardError",
   ExportString[
      First[
         Cases[
            Import[#,"XML"],
            XMLElement["string",_,{url_}]:>url,
            Infinity
         ]
      ]&/@FileNames[],
      "Text"
   ]
]

So, here is my search history through the web in order to find out why Stoustrups code didn't work:

https://duckduckgo.com/?q=21st+century+bjarne+stoustrup+linkedin&t=osx
https://duckduckgo.com/?q=21st+century+bjarne+stoustrup+linkedin&t=osx&ia=web
https://duckduckgo.com/?q=21st+Century+C%2B%2B&t=osx
https://duckduckgo.com/?q=21st+Century+C%2B%2B&t=osx&ia=web
https://news.ycombinator.com/item?id=42946321
https://cacm.acm.org/blogcacm/21st-century-c/
https://cacm.acm.org/blogcacm/21st-century-c/#refer
https://stroustrup.com/index.html
https://en.cppreference.com/w/cpp/compiler_support
https://stroustrup.com/compilers.html
https://stackoverflow.com/questions/62765630/how-to-use-c20-modules-with-gcc
https://duckduckgo.com/?q=clang+C%2B%2B23&t=osx
https://duckduckgo.com/?q=clang+C%2B%2B23&t=osx&ia=web
https://clang.llvm.org/
https://clang.llvm.org/cxx_status.html
https://clang.llvm.org/features.html
https://duckduckgo.com/?q=c%2B%2B+standard&t=osx
https://duckduckgo.com/?q=c%2B%2B+standard&t=osx&ia=web
https://duckduckgo.com/?q=c%2B%2B+use+import+std&t=osx
https://duckduckgo.com/?q=c%2B%2B+use+import+std&t=osx&ia=web
https://duckduckgo.com/?q=extension+af+c%2B%2B+program+files&t=osx
https://duckduckgo.com/?q=extension+af+c%2B%2B+program+files&t=osx&ia=web
https://en.cppreference.com/w/cpp/feature_test#cpp_lib_modules
https://stackoverflow.com/questions/1545080/c-code-file-extension-what-is-the-difference-between-cc-and-cpp
https://duckduckgo.com/?q=gcc+g%2B%2B+use+modules&t=osx
https://duckduckgo.com/?q=gcc+g%2B%2B+use+modules&t=osx&ia=web
https://www.google.com/?client=safari
https://duckduckgo.com/?q=hos+to+use+c%2B%2B23+in+clang&t=osx
https://duckduckgo.com/?q=hos+to+use+c%2B%2B23+in+clang&t=osx&ia=web
https://duckduckgo.com/?q=how+to+use+import+instead+of+%23include+in+modern+c%2B%2B&t=osx
https://duckduckgo.com/?q=how+to+use+import+instead+of+%23include+in+modern+c%2B%2B&t=osx&ia=web
http://clang.llvm.org/
https://isocpp.org/std/the-standard
https://learn.microsoft.com/en-us/cpp/cpp/tutorial-import-stl-named-module?view=msvc-170
https://duckduckgo.com/?q=unique-lines.cpp%3A1%3A1%3A+error%3A+unknown+type+name+%27import%27&t=osx
https://duckduckgo.com/?q=which+c%2B%2B+compilers+support+modules&t=osx
https://duckduckgo.com/?q=which+c%2B%2B+compilers+support+modules&t=osx&ia=web
https://duckduckgo.com/?q=which+compiler+use+stroustrup&t=osx
https://duckduckgo.com/?q=which+compiler+use+stroustrup&t=osx&ia=web
https://www.reddit.com/r/cpp/comments/ok8o4t/why_is_there_still_such_poor_compiler_support_for/

Back to the main story

The first line in Stoustrups C++ code utilises C++ modules, that, even though they became part of the C++ standard in 2020, are not yet implemented in the two main C++ compilers on my laptop: Apple's clang++ and g++. At least that's what I inferred from a few experiments, readings of the --help output, and from the web.

So, one might say, that it's very easy for Stoustrup to complain and worry about why people still write old-fashioned C++ code, when the "new" functionality isn't available yet...

I then did one more search, and ended on Stackoverflow: apple-clang-13-c20-module-support-missing and decided to install yet another C++ compiler: brew install llvm. Then, with a bit of help from ChatGPT, I changed Stoustrups example to this:

import <iostream>;
import <unordered_map>  // Explicitly import unordered_map
using namespace std;

/*
  dette er linje 1
  dette er linje 1
  dette er linje 1
  dette er linje 2
*/

int main()
{
  unordered_map<string,int> m;
  for (string line; getline(cin,line);)
    if (m[line]++ == 0)
      cout << line << '\n';
}

compiled it with the newly installed Homebrew version of llvm

clang++ -std=c++2a -fmodules -o unique-lines unique-lines.cpp

and got the correct result:

>: ./unique-lines < unique-lines.cpp
import <iostream>;
import <unordered_map>;  // Explicitly import unordered_map
using namespace std;

/*
  dette er linje 1
  dette er linje 2
*/
int main()
{
  unordered_map<string,int> m;
  for (string line; getline(cin,line);)
    if (m[line]++ == 0)
      cout << line << '\n';
}

Now, it's time to read the rest of Stoustrups article to maybe learn why import std; doesn't work, but import <iostream> and import <unordered_map>; is necessary. Oh, I also just wanted to test whether the awk code also works as stated:

 >: awk '!a[$0]++' < unique-lines.cpp 
import <iostream>;
import <unordered_map>;  // Explicitly import unordered_map
using namespace std;

/*
  dette er linje 1
  dette er linje 2
*/
int main()
{
  unordered_map<string,int> m;
  for (string line; getline(cin,line);)
    if (m[line]++ == 0)
      cout << line << '\n';
}

Yup! correct result. Hooray for old school 😃

Bibliography

• Published as 21st Century C++ – Communications of the ACM. (2025, February 4), but also by Stoustrup in PDF format as 21st Century C++, Columbia University, Bjarne Stroustrup