Searching inside PDF works for me, but there may be some room to improve it. For example, if I select a paragraph inside PDF reader and paste it out and some phrase (e.g. "spherical aberration") are broken into two lines, then the search of such phrase will fail. Although individually they can be found, but search of "spherical" by itself is too broad to be useful. BTW, the phase search inside a PDF view will also fail. However, I accidentally found out that if I attach the PDF in Evernote, its search engine could find the phrase.

So I wonder if it is possible for Zotero to join the extracted text from PDF automatically so that such phase search could be working. Thanks!